Changes of Flow and Sediment Transport in the Lower Min River in Southeastern China Under the Impacts of Climate Variability and Human Activities

water

Article Changes of Flow and Sediment Transport in the Lower Min River in Southeastern China under the Impacts of Climate Variability and Human Activities

Wen Wang 1,*, Tianyue Wang 1, Wei Cui 1,2, Ying Yao 1, Fuming Ma 3, Benyue Chen 3 and Jing Wu 3

1 State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering Sciences, Hohai University, Nanjing 210000, China; [email protected] (T.W.); [email protected] (W.C.); [email protected] (Y.Y.) 2 Nanjing Hydraulic Research Institute, Nanjing 210000, China 3 Bureau of Hydrology and Water Resources of Fujian Province, Fuzhou 350000, China; [email protected] (F.M.); [email protected] (B.C.); [email protected] (J.W.) * Correspondence: [email protected]

Abstract: The Min River is the largest river in Fujian Province in southeastern China. The construction of a series of dams along the upper reaches of the Min River, especially the Shuikou Dam, which started filling in 1993, modified the flow processes at the lower Min River, leading to the significant increase in low-flows and slightly decrease in flood-flows. At the same time, reservoirs have more effects on the sediment transport process than flow process by trapping most sediment in the reservoirs, and greatly reduced the amount of sediment transporting downstream. Increase in vegetation cover also contributes to the decrease in sediment yield. The reduction in sediment together with excessive sand mining in the lower Min River resulted in the severe downward erosion Citation: Wang, W.; Wang, T.; Cui, of the riverbed. Using a reformulated elasticity approach to quantifying climatic and anthropogenic W.; Yao, Y.; Ma, F.; Chen, B.; Wu, J. Changes of Flow and Sediment contributions to sediment changes, the relative contribution of precipitation variability and human Transport in the Lower Min River in activities to sediment reduction in the lower Min River are quantified, which shows that the sed- Southeastern China under the iment reduction is fully caused by human activities (including land use/land cover changes and Impacts of Climate Variability and dam construction). Human Activities. Water 2021, 13, 673. https://doi.org/10.3390/ Keywords: anthropogenic impact; hydrologic change; dam construction; suspended sediment; w13050673 streamflow; precipitation; vegetation; elasticity approach

Academic Editor: Silvia Kohnová

Received: 31 December 2020 1. Introduction Accepted: 26 February 2021 Published: 2 March 2021 It is known that the construction of reservoir dams will change the timing, magnitude, and frequency of low and high ﬂows (Magilligan and Nislow, 2005; Wang et al., 2011) [1,2].

Publisher’s Note: MDPI stays neutral In addition, the sediment transported to downstream is reduced after dam construction. with regard to jurisdictional claims in Globally, about 50% of the gross sediment flux was estimated being trapped in reservoirs published maps and institutional affil- (Nilsson et al., 2005) [3]. On an average, annual sediment fluxes of major large rivers iations. in East, Southern, and Southeast Asian region reduced by more than 75% from 1960s to 2000s due to the effect of mega dams (Gupta et al., 2012) [4]. After the construction of the Three Gorges Dam (TGD) in China in 2003, the sediment discharge in the lower reaches of the Yangtze River decreased by 55% compared with 1993–2002, and 65% of the pre- to post-TGD decrease in sediment flux can be attributed to the TGD (Yang et al., Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. 2015) [5]. Due to the effect of sediment trapping in the Lancang cascade dams, the sediment This article is an open access article reduced by 74.1% in the Vietnamese Mekong Delta over a 55-year period (1961–2015) distributed under the terms and (Binh et al., 2020) [6]. In addition to the effects of dam construction, precipitation changes conditions of the Creative Commons and vegetation changes resulted from climate changes (including changes of precipitation Attribution (CC BY) license (https:// and temperature) may also have significant effects on runoff and sediment yield in the creativecommons.org/licenses/by/ watershed (Song et al., 2016; Zhang et al., 2019) [7,8]. The reduction in sediment load 4.0/). would lead to riverbed erosion (Binh et al., 2020; Zheng et al., 2018) [6,9] river channel

Water 2021, 13, 673. https://doi.org/10.3390/w13050673 https://www.mdpi.com/journal/water Water 2021, 13, 673 2 of 19

adjustment (Xia et al., 2016) [10], salt tide intrusion in estuaries (Eslami et al., 2019) [11], and coastal area loss in river deltas (Luo et al., 2017; Liu et al., 2017) [12,13]. On the other hand, while all dams affect the hydrologic regime, the magnitude and type of impacts vary greatly by dam and location (Timpe and Kaplan, 2017) [14]. To quantify dam-induced hydrological impacts and quantify the relative contribution of climate and human effects to hydrological alteration across different geophysical regions would be helpful to build up a holistic view of human effects on hydrological dynamics and consequently helpful for sustainable sediment manage and future dam siting and operation. The Min River is the largest river in Fujian Province of China with a series of major reservoirs built in the catchment, among which Shuikou Dam is the largest one. This reservoir controlled 86% of the drainage area of the Min River. In the present study, we will analyze how the hydrology in the lower reaches of the Min River, including the streamﬂow and the sediment transport, is altered under the joint effects of a series of dams, especially the construction of Shuikou Dam, and quantify the relative contribution of different factors to sediment load changes.

2. Data and Methods 2.1. Introduction about the Min River Basin The Min River has a total length of 541 km and a total drainage area of 60,992 km2. It originates from the Wuyi mountain range, flows from the west to the east, and pours into the East China Sea. The basin is mostly mountainous (Figure1), and the average slope of the river is 5‰. The basin is dominated by the subtropical monsoon climate zone with a warm and humid climate. The average annual temperature is about 17–19 ◦C. The average annual precipitation is about 1600–1700 mm, declining gradually from upstream to downstream. The period during April to September is the flood season, and the runoff takes up about 75% of the total annual runoff. The maximum monthly runoff occurs in June when the main flood season in the river basin starts. From October to March of the following year, it is a low-flow season, and the average annual runoff reaches the minimum in January. The lower reaches of the Min River are often influenced by heavy typhoon rains from July to September, causing serious floods with high flood peaks and large water volume. The Min River basin has very good vegetation coverage generally. According to GlobeLand30 land cover data at a 30 m resolution provided by National Geomatics Center of China (globallandcover.com), 74.5% of the basin is covered by forest. As a result, the sediment yield is very low, mostly below 0.14 kg/m3 annually on average in the tributaries of Min River. Several islands are present in the lower Min River channel near Fuzhou city. The river estuary is strongly affected by normal semi-diurnal tides. The maximum tidal range in the estuary is 7.04 m, and the average tidal range is 4.1 m. Nine large reservoirs have been built in the Min basin, including Ansha Reservoir, Chi- tan Reservoir, Dongxi Reservoir, Gutianxi Reservoir, Shaxikou Reservoir, Shuikou Reservoir, Shuidong Reservoir, Jiemian Reservoir, and Jinzhong Reservoir (see Figure1). Before 1970, only one major reservoir was built in 1959. During 1971–1990, 4 major reservoirs were completed. After 1990, four more reservoirs were completed. Among them, the Shuikou Reservoir, which is located in about 120 km above the Min River estuary, is the largest one. The construction of Shuikou Dam started in March 1987, and the filling of the reservoir began in March 1993. The reservoir has a maximum storage capacity of 2.6 billion m3 and an effective storage capacity of 700 million m3. The drainage area above the dam site is 52,438 km2, accounting for 86% of the total basin area. The reservoir is mainly used for hydroelectricity, also used for flood control and navigation. Water 2021, 13, 673 3 of 19 Water 2021, 13, x FOR PEER REVIEW 3 of 20

Dongxi Reservoir (1986) Total capacity: 113 million m3

Shaxikou Reservoir (1984) Chitan Reservoir (1982) 3 3 Total capacity:164 million m Total capacity:870 million m YangKou • Gutianxi Reservoir (1959) Total capacity:647 million m3 Shuikou Reservoir (1993) Shuidong Reservoir (1995) Total capacity:2600 million m3 Ansha Reservoir (1987) Total capacity:108 million m3 Total capacity:735 million Xiapu 3 m Fuzhou Zhuqi Liberation Bridge Wenshanli Baiyantan

Jiemian Reservoir (2010) Total capacity:1824 million m3

Jinzhong Reservoir (2010) Total capacity:106 million m3

FigureFigure 1. TheThe elevation elevation of of the the Min Min River River basin basin and locations of major reservoirs and hy hydrologicdrologic gauging sites.

2.2. DataThe UsedMin River basin has very good vegetation coverage generally. According to GlobeLand30Major hydrologic land cover gauging data at a stations 30 m resolution located along provided the lower by National reaches Geomatics of the Min Center River of(below China the (globallandcover.com), Shuikou Dam) are displayed 74.5% of in Figurethe basin1. In is our covered study, by the forest. following As a hydrologic result, the 3 sedimentdata are used: yield is very low, mostly below 0.14 kg/m annually on average in the tributaries of Min River. (1) Daily discharge data at Zhuqi station during 1950–2017. Zhuqi is the most important Several islands are present in the lower Min River channel near Fuzhou city. The station along the lower reaches, located at about 45 km below the Shuikou Dam. It river estuary is strongly affected by normal semi-diurnal tides. The maximum tidal range controls a drainage area of about 54,500 km2. The observation of water level and river in the estuary is 7.04 m, and the average tidal range is 4.1 m. discharge at Zhuqi started in 1950; Nine large reservoirs have been built in the Min basin, including Ansha Reservoir, (2) Daily water level data at Xiapu, Zhuqi, Wenshanli, Liberation Bridge, and Baiyantan Chitan Reservoir, Dongxi Reservoir, Gutianxi Reservoir, Shaxikou Reservoir, Shuikou from 1950 to 2017 (except Xiapu, which started in 1993); Reservoir, Shuidong Reservoir, Jiemian Reservoir, and Jinzhong Reservoir (see Figure 1). (3) Daily suspended sediment concentration data at Zhuqi during 1952–2017; Before 1970, only one major reservoir was built in 1959. During 1971–1990, 4 major (4) Daily precipitation data at 19 precipitation gauging sites during 1950 (or 1951 or 1952) reservoirsto 2017. were The completed. daily precipitation After 1990, data four are convertedmore reservoirs to the arealwere precipitationcompleted. Among in Min them,River the Shuikou basin by Rese the Thiessonrvoir, which polygon is located method. in about 120 km above the Min River estuary, is the largest one. The construction of Shuikou Dam started in March 1987, and The bathymetric data of the river channel along the cross-section at the Zhuqi in 2000, the filling of the reservoir began in March 1993. The reservoir has a maximum storage 2003, 2008, 2011, and 2014, and the thalweg data, which are derived from the lowest points capacity of 2.6 billion m3 and an effective storage capacity of 700 million m3. The drainage of 79 cross-sections (about every 2 km along the main channel below Shuikou Dam) in 2011 area above the dam site is 52,438 km2, accounting for 86% of the total basin area. The and 2015 are used to analyze the change of river channel under the impact of Shuikou Dam reservoir is mainly used for hydroelectricity, also used for flood control and navigation. operation. All the bathymetric data are provided by the Bureau of Hydrology and Water Resources of Fujian Province.

Water 2021, 13, 673 4 of 19

In addition, the GIMMS (Global Inventory Monitoring and Modeling System) third- generation NDVI (Normalized Difference Vegetation Index) data during 1981–2015 released by NASA (https://glam1.gsfc.nasa.gov) is used for analyzing the change of vegetation in the Min River basin.

2.3. Methods 2.3.1. Indicators of Flow Regime Changes Richter et al. (1996) [15] proposed the RVA (Range of Variability Approach) method to evaluate the extent to which streamflow changed by human perturbations. They defined a suite of 32 ecologically relevant IHA (Indicators of Hydrological Alteration) parameters regarding the magnitude, timing, duration, frequency, and rate of water condition changes. In the present study, we investigate variations in hydrological conditions of not only the flow process (with focus on flood flows and low flows) but also the sediment transport process, therefore the following 10 indices are used to measure changes in the magnitude, timing, duration and frequency of flow and sediment conditions: • The mean annual discharges (MAD) • The annual coefficient of variation of daily discharges (CV) • The annual minimum 7-day average discharge (Min7d) • The annual maximum 1-day average discharge (Max1d) • Julian date of each annual 1-day maximum (Dmax1d) • Occurrence day of each annual 7-day minimum starting from August 1 (Dmin7d) • The number of days with the discharge below the 15% percentile (N15p) • The number of days with the discharge exceeding the 90% percentile (N90p) • The mean annual concentration of suspended sediment (MSSC) • The annual total load of suspended sediment (TSSL)

2.3.2. Methods of Change Detection (1) Exploratory change analysis The double mass curve method (Searcy et al., 1960) [16] is commonly used to check the consistency of the relationship among different kinds of hydrologic data by comparing one series of data with another in the area. A double mass curve is a scatterplot of the cumulative sums of two variables. It exhibits a straight line when the relationship between the two keeps unchanged, and a break in the slope of the double mass curve indicates the change of their relationship. (2) Mann-Kendall trend test The Mann-Kendall trend test (Kendall, 1975) [17], referred to as MK test hereafter, is used to test for the existence of possible trends in vegetation cover, precipitation, ﬂow, and sediment transport processes. MK test is a rank-based nonparametric method, which is less sensitive to outliers than other parametric statistics. The MK method can test trends in a time series without specifying whether the trend is linear or nonlinear. We identify the type of temporal changes according to Kendall’s τ, which measures the strength of the monotonic trend, jointly with the p-value, which measures the level of signiﬁcance, shown in Table1.

Table 1. Classiﬁcation of trend in terms of p-value and τ with Mann-Kendall (MK) trend test.

τ p-Value Class Trend Type p < 0.01 +3 Very significant increase τ > 0 0.01 ≤ p < 0.05 +2 Significant increase 0.05 < p ≤ 0.1 +1 Slight increase τ = 0 0.1 < p 0 No trend 0.05 < p ≤ 0.1 −1 Slight decrease τ < 0 0.01 ≤ p < 0.05 −2 Significant decrease p < 0.01 −3 Very significant decrease Water 2021, 13, 673 5 of 19

(3) Pettitt test for change-point detection The approach after Pettitt (1979) [18] is commonly applied to detect a single change- point in hydrological series or climate series. It tests the H0: no change, against the alternative: a change point exists. The non-parametric statistic is deﬁned as:

KT = max|Ut,T|, 1 ≤ t ≤ T, (1)

where the statistic Ut,T is equivalent to a Mann–Whitney statistic for testing that the two samples X1,..., Xt and Xt+1,..., XT come from the same population, given by:

t T Ut,T = ∑ ∑ sgn Xi − Xj , t = 2, 3, . . . , T, (2) i=1 j=t+1

where sgn(x) = 1 if x > 0, 0 if x = 0, and −1 if x < 0. The signiﬁcance probability associated with the value KT is approximately given by:

n 2 3 2o p ≈ 2 exp −6Kt,T/ T + T . (3)

The change-point of the series is located at KT, provided that the statistic KT is signiﬁ- cant at a given signiﬁcance level (e.g., α = 0.05).

2.3.3. Quantification of the Attribution of Sediment Discharge Change Generally, changes of suspended sediment load (SSL) can be attributed to climate variability and human activities. The elasticity approach proposed by Zhang et al. (2019) [19] is adopted and reformulated to quantify the effects of climate variability and human activities on changes of suspended sediment discharge in different periods, which are identified with abrupt-change detection methods. Because the amount of sediment transport (QS) is the multiplication of the concentration of sediment (C) and streamflow discharge (Q), changes in QS can be expressed in an elasticity form as (Zhang et al., 2019) [19]:

dQS dQ dC = ηQ + ηC , (4) QS Q C

∂QS/QS ∂QS/QS where ηQ = ∂Q/Q and ηC = ∂C/C denote the elasticity of SSL to the streamflow and sediment concentration ~ streamflow (C–Q) relationships, respectively. The reduction in Qs between two periods can be divided into two parts, i.e., the part due to changes in streamflow (∆Qs,Q) and the part due to changes in the C–Q (∆Qs,C), which are given by n ∆Qs,Q = ∑ qjCj,1 pj,2 − pj,1 (5) j=1 n ∆Qs,C = ∑ qj pj,2 Cj,2 − Cj,1 (6) j=1

where qj is the average daily discharge in class j (j = 1, ... , n), which is one of the n non-overlapping discharge classes with equal length of 1/n of the range of streamﬂow variation in log-space; pj,1 and pj,2 are the occurrence probability of discharge qj in two periods (i.e., period 1 and period 2) when the C~Q relationship changed; Cj,1 and Cj,2 are the sediment concentrations for discharge pj in period 1 and period 2, respectively; QS is the average daily SSL; Q1, Q2, and Q denote mean values of daily discharge in period 1, period 2, and the whole period, respectively; C1, C2, and C denote mean values of sediment concentration in period 1, period 2, and the whole period, respectively. Water 2021, 13, 673 6 of 19

Therefore, we have h i ∑n q C p − p /Q ∂QS/QS j=1 j j,1 j,2 j,1 S ηQ = ≈ (7) ∂Q/Q Q2 − Q1 /Q

h i ∑n q p C − C /Q ∂QS/QS j=1 j j,2 j,2 j,1 S ηC = ≈ . (8) ∂C/C C2 − C1 /C The relative change in Q due to changes in climate (i.e., precipitation P, actual evaporation E, and potential evaporation E0) and catchment property (m) is given by (Michael and Farquhar, 2011) [20]:

dQ P ∂E dP E ∂E dE m ∂E dm = 1 − − 0 0 − , (9) Q Q ∂P P Q ∂E0 E0 Q ∂m m

m m ∂E E E0 ∂E E P with the respective partial differentials given by = m m , = m m and ∂P P P +E0 ∂E0 E0 P +E0 m m m m ∂E E ln(P +E0 ) P ln P+E0 ln E0 = − m m . The catchment property (m) can be estimated based ∂m m m P +E0 on the catchment water balance equation and Budyko equation [21], which is described by:

Q = P − E − ∆W (10)

PE E = 0 (11) m m1/m P + E0 where ∆W is the change in storage within the catchment which can be ignored in the long-term; E0 is estimated using the FAO-56 method [22]. By combing Equations (10) and (11), we can estimate m by solving the nonlinear equation. By combining Equations (4) and (9), we have a comprehensive elasticity form for QS change given as dQS dP dE0 dm dC = ηP + ηE0 + ηm + ηC , (12) QS P E0 m C h i h i P ∂E E0 ∂E with ηP = η 1 − and ηE = −η , denoting the elasticity of SSL to stream- Q Q ∂P 0 Q Q ∂E0 h m ∂E i ﬂow caused by variability of precipitation and E0, respectively, and ηm = −ηQ Q ∂m , denoting the elasticity of SSL to streamﬂow caused by changes of catchment property (such as topography, soil type/depth, geologic substrate, land cover, etc.). Here, ηP and

ηE0 represent the effects of climate variability on SSL change, while ηm and ηC represent mostly the effects of human activities. Changes in SSL attributed to changes in precipitation (∆Qs,P), potential evapotranspi-

ration (∆Qs,E0 ), and land surface (by changing streamﬂow, ∆Qs,m, and moderating C~Q relationships, ∆Qs,C) can be, respectively, expressed as (Zhang et al. 2019) [19]:

∆P ∆E0 ∆m ∆C ∆Qs,P = ηp Qs, ∆Qs,E0 = ηE0 Qs, ∆Qs,m = ηm Qs, ∆Qs,C = ηC Qs (13) P E0 m C

The sum of ∆Qs,P and ∆Qs,E0 represents the contribution of climate variability to changes of sediment discharge, whereas the sum of ∆Qs,m and ∆Qs,C represents the contribution of human activities (including land use/cover change, dam construction, etc.). Therefore, the relative contribution of climate change and human activity are estimated by:

εs,P = ∆Qs,P/∆Qs, εs,E0 = ∆Qs,E0 /∆Qs, εs,m = ∆Qs,m/∆Qs, εs,C = ∆Qs,C/∆Qs (14)

where ∆Qs = ∆Qs,P + ∆Qs,E0 + ∆Qs,m + ∆Qs,C. Water 2021, 13, x FOR PEER REVIEW 7 of 20

moderating C~Q relationships, ∆𝑄, ) can be, respectively, expressed as (Zhang et al. 2019) [19]:

∆𝑃 ∆𝐸 ∆𝑚 ∆𝐶 ∆𝑄, =𝜂 𝑄,∆𝑄, =𝜂 𝑄,∆𝑄, =𝜂 𝑄,∆𝑄, =𝜂 𝑄 (13) 𝑃 𝐸 𝑚 𝐶

The sum of ∆𝑄, and ∆𝑄, represents the contribution of climate variability to changes of sediment discharge, whereas the sum of ∆𝑄, and ∆𝑄, represents the contribution of human activities (including land use/cover change, dam construction, etc.). Therefore, the relative contribution of climate change and human activity are estimated by: 𝜀 =∆𝑄 /∆𝑄 ,𝜀 =∆𝑄 /∆𝑄 ,𝜀 =∆𝑄 /∆𝑄 ,𝜀 =∆𝑄 /∆𝑄 Water 2021, 13, 673 , , , , , , , , 7(14) of 19

where ∆𝑄 =∆𝑄, +∆𝑄, +∆𝑄, +∆𝑄,.

3. Alteration of Streamflow and Sediment Process in the Lower Reaches of the Min 3. Alteration of Streamflow and Sediment Process in the Lower Reaches of the RiverMin River 3.1.3.1. Changes Changes of of Streamflow Streamflow Process Process TheThe mean mean annual annual discharges discharges (MAD) (MAD) observed observed at at Zhuqi Zhuqi during during 1950 1950 to to 2017 2017 are are plottedplotted in in Figure Figure 22a.a. The average annual discharge isis 17181718 mm3/s/s during during the the whole whole 68 68 years, years, 16851685 m m33/s/s during during 1950–1992, 1950–1992, and and 1775 1775 m m33/s/s during during 1993–2017. 1993–2017. Figu Figurere 2a2a shows shows no no obvious obvious overalloverall trend trend during during the the whole whole 68 68 years. years. In In co comparison,mparison, the the annual annual coefficient coefficient of of variation variation (CV)(CV) of of daily daily discharges discharges exhibits exhibits a a signific significantant downward downward trend, trend, as as shown shown in Figure 22b.b. CV 0.60.81.01.21.41.61.8

1950 1960 1970 1980 1990 2000 2010 Year (a) (b)

FigureFigure 2. 2. MeanMean annual annual discharge discharge (a (a) )and and annual annual coefficient coefﬁcient of of variation variation (CV) (CV) of of daily daily discharge discharge (b ()b ) observedobserved at at Zhuqi Zhuqi during during 1950–2017. 1950–2017.

TheThe changes changes of of low low flows flows and and flood flood flows flows during during 1950~2017 1950~2017 are are further further investigated. investigated. TheThe annual annual series series of of minimum minimum 7-day 7-day average average discharge discharge (Min7d), (Min7d), maximum maximum 1-day 1-day average average dischargedischarge (Max1d), (Max1d), minimum minimum 7-day 7-day average average flow flow occurrence occurrence time time (Dmin7d), (Dmin7d), maximum maximum 1-day1-day average average flow flow occurrence occurrence time time (Dmax1d) (Dmax1d),, the the number number of of days days with with the the discharge discharge belowbelow the the 15% 15% percentile percentile (~510 (~510 m m33/s)/s) (N15p), (N15p), and and the the number number of of days days with with the the discharge discharge exceedingexceeding the the 90% 90% percentile percentile (~3470 (~3470 m m33/s)/s) (N90p) (N90p) are are displayed displayed in in Figure Figure 33.. By visualvisual inspection,inspection, we we find find an an obvious obvious positive positive tren trendd in in Min7d Min7d and and an an obvious obvious negative negative trend trend in in Max1d,Max1d, at at the the same same time, time, both both N15p N15p and and N90p N90p decreased. decreased. While While other other sequences sequences showed showed nono obvious obvious changing changing trend. trend. Mann-Kendall trend test is applied to the annual flow, low flow, and flood flow series at Zhuqi. The test results are presented in Table2, which show that MAD exhibits no significant changes, but significant negative trend is observed in the CV series; Min7D significantly increased, while the N15p significantly decreased; Max1D did not changed significantly although N90p significantly decreased. That is, the regulation of the reservoirs has no effect on total runoff although the increase in water surface due to the impounding of reservoirs may increase the evaporation slightly; reservoirs have significant impacts on low flows, but minor effects on flood flows. Overall, the regulation of reservoirs makes the runoff tend to be slightly more evenly distributed over the year, but because the effective capacity of reservoirs is far less than the total annual runoff, so the impacts on flood peak flow is not remarkable. Table2 also presents the results of Pettitt test, which indicate that some flow indices of Min River, including CV, Min7D, N15p, and N90p, experienced significant step changes in early 1980s’. WaterWater 20212021, 13, 13, x, 673FOR PEER REVIEW 8 of8 of 20 19

FigureFigure 3. 3.ChangesChanges of of low low flows flows and and flood flood flows flows observedobserved atat ZhuqiZhuqi duringduring 1950 1950 to to 2017. 2017. (a ()a The) The annual annualseries series of minimum of minimum 7-day 7-da averagey average discharge discharge (Min7d); (Min7d); (b) minimum (b) minimum 7-day 7-day average average flow occurrenceflow occurrencetime (Dmin7d); time (Dmin7d); (c) the number (c) the ofnumber days withof days the with discharge the discharge below the below 15% the percentile 15% percentile (~510 m 3/s) (~510(N15p); m3/s) (d (N15p);) maximum (d) maximum 1-day average 1-day discharge average (Max1d); discharge (e )(Max1d); maximum (e) 1-day maximum average 1-day flow average occurrence flowtime occurrence (Dmax1d); time (f) the (Dmax1d); number of(f) days the number with the of discharge days with exceeding the discharge the 90% exceeding percentile the (~3470 90% m3/s) 3 percentile(N90p). (~3470 m /s) (N90p).

TableMann-Kendall 2. Trend and change-points trend test is applied in flow series to the detected annual byflow, the MKlow testflow, and and Pettitt flood test. flow series at Zhuqi. The test results are presented in Table 2, which show that MAD exhibits no significantData changes, Type but significant Index negative trend is observed Trend in the CV Year series; of Change Min7D significantly increased, while theMAD N15p significantly decreased; 0 Max1D did not – changed significantly although N90p significantlyCV decreased.− That** is, the regulation 1984 ** of the reservoirs has no effect on total Min7drunoff although the increase + ** in water surface 1981 due ** to the Dmin7d) 0 – impoundingFlow of reservoirs may increase the evaporation slightly; reservoirs have significant impacts on low flows,N15p but minor effects on flood− ** flows. Overall, the 1980 regulation ** Max1d 0 – of reservoirs makes the runoff Dmax1dtend to be slightly more evenly 0 distributed over – the year, but because the effective capacityN90p of reservoirs is far less− **than the total annual 1984 runoff, ** so the impacts on flood peak flow isMSSC not remarkable. − ** 1993 ** Sediment TSSL − ** 1993 ** TableNote: 2. + Trend indicates and positive change-points trend; − indicatesin flow series negative detected trend; **by indicates the MK that test theand change Pettitt is test. significant at 0.01 significance level. Data Type Index Trend Year of Change The seasonal distribution ofMAD flow processes under0 conditions of different – degree of reservoir impacts is investigatedCV by comparing the average− ** discharges in each1984 ** day of the Min7d + ** 1981 ** year (smoothed with 8 FourierDmin7d) harmonics) during three0 different periods, i.e., – 1950–1970, Flow 1971–1992, and 1993–2017, in FigureN15p4 a. For comparison,− the** variation of average 1980 ** precipitation in each day of the year duringMax1d the three periods is0 plotted in Figure4b. – Comparing Figure4a,b, the seasonal distributionDmax1d of flow basically matches0 that of precipitation – and did not changed much in the three periodsN90p by visual inspection.− ** It is known that 1984 reservoirs ** gen- MSSC − ** 1993 ** erally controlSediment the water storage below a certain level before the major flood season (April TSSL − ** 1993 ** to July in Min River basin) to prepare for the occurrence of major floods, then stores water Note: + indicates positive trend; − indicates negative trend; ** indicates that the change is significantafter the at major 0.01 significance flood season, level. and finally increases the release of outflow in the low-flow period. However, because the effective storage capacity of all the 8 major reservoirs above Zhuqi station takes about only 7% of the average annual runoff of the Min River observed at Zhuqi, the impacts on flow processes are not obvious for high flows, but significant for low flows during November to February.

Water 2021, 13, x FOR PEER REVIEW 9 of 20

Table 2 also presents the results of Pettitt test, which indicate that some flow indices of Min River, including CV, Min7D, N15p, and N90p, experienced significant step changes in early 1980s’. The seasonal distribution of flow processes under conditions of different degree of reservoir impacts is investigated by comparing the average discharges in each day of the year (smoothed with 8 Fourier harmonics) during three different periods, i.e., 1950–1970, 1971–1992, and 1993–2017, in Figure 4a. For comparison, the variation of average precipitation in each day of the year during the three periods is plotted in Figure 4b. Comparing Figure 4a,b, the seasonal distribution of flow basically matches that of precipitation and did not changed much in the three periods by visual inspection. It is known that reservoirs generally control the water storage below a certain level before the major flood season (April to July in Min River basin) to prepare for the occurrence of major floods, then stores water after the major flood season, and finally increases the release of outflow in the low-flow period. However, because the effective storage capacity of all the 8 major reservoirs above Zhuqi station takes about only 7% of the average annual runoff Water 2021, 13, 673 9 of 19 of the Min River observed at Zhuqi, the impacts on flow processes are not obvious for high flows, but significant for low flows during November to February.

1950-1970 1950-1970 1971-1992 1971-1992 1994-2017 1994-2017 Discharge(cms) Precipitation (mm) 1000 2000 3000 4000 5000 024681012

0 100 200 300 0 100 200 300 Day over the year Day over the year (a) (b)

FigureFigure 4. 4. (a)(a )Variation Variation of of areal areal precipitation precipitation above above the the drainage drainage area area above above Zhuqi Zhuqi and and (b ()b average) average dailydaily discharge discharge observed observed at at Zhuqi Zhuqi during during three three different different periods. periods.

3.2.3.2. Variation Variation in in Sediment Sediment Transport Transport TheThe annual annual average average concentration concentration of of susp suspendedended sediment sediment (SSC) (SSC) observed observed at at Zhuqi Zhuqi duringduring 1952 1952 to to 2017 2017 are are plotted plotted in in Figure Figure 5.5. According According to to the the observation observation at at Zhuqi, Zhuqi, before before thethe construction construction of of Shuikou Shuikou Dam, Dam, the the average average suspended suspended sediment sediment concentration concentration was was 3 3 0.1290.129 kg/m kg/m3 duringduring the the period period from from 1952 1952 to to 1992 1992 with with a a maximum maximum of of 0.261 0.261 kg/m 3 inin 1962 1962 3 andand the the minimum minimum of of 0.065 0.065 kg/m kg/m3 inin 1991, 1991, and and the the average average suspended suspended sediment sediment transport transport capacitycapacity was was 7.15 7.15 million million tons tons every every year year during during 1951 1951 to to1992. 1992. After After the theimpoundment impoundment of Shuikouof Shuikou Dam Dam in 1993, in 1993, the average the average suspended suspended sediment sediment concentration concentration at Zhuqi at Zhuqiwas 0.038 was 3 3 kg/m0.0383 during kg/m 1993–2017,during 1993–2017, with a maximum with a maximum of 0.136 of kg/m 0.1363 in kg/m 2010 andin 2010 a minimum and a minimum of less 3 thanof less 0.007 than kg/m 0.0073 in kg/m2008, andin 2008,the average and the annual average suspended annual suspended sediment transport sediment decreased transport todecreased 2.48 million to 2.48tonsmillion during tons1993 duringto 2017. 1993 In comparison to 2017. In with comparison the average with of the 7.15 average million of 7.15 million tons every year during 1951 to 1992, about 4.67 million tons (about 65%) tons every year during 1951 to 1992, about 4.67 million tons (about 65%) suspended suspended sediment was deposited in all the reservoirs above Zhuqi every year. By sediment was deposited in all the reservoirs above Zhuqi every year. By comparing the comparing the average of 6.06 million tons of sediment transport during the more recent average of 6.06 million tons of sediment transport during the more recent period from period from 1983 to 1992 with the 4.67 million tons of sediment after 1993 when Shuikou 1983 to 1992 with the 4.67 million tons of sediment after 1993 when Shuikou Reservoir Reservoir was built, it is estimated that about 3.58 million tons of suspended sediment, Water 2021, 13, x FOR PEER REVIEW was built, it is estimated that about 3.58 million tons of suspended sediment, which would10 of 20 which would deposit in the lower Min River channel without the effect of reservoirs, was deposit in the lower Min River channel without the effect of reservoirs, was deposited in deposited in the Shuikou reservoir every year. the Shuikou reservoir every year.

Sediment load (million ton) (million load Sediment 0 5 10 15 20

1950 1960 1970 1980 1990 2000 2010 Year (a) (b)

FigureFigure 5. 5. (a(a) )Annual Annual average average suspended suspended se sedimentdiment concentration andand ((bb)) annualannual suspendedsuspended sediment sedimentload observed load observed at Zhuqi at during Zhuqi 1952–2017. during 1952–2017.

AccordingAccording to to the the measurement measurement at at Zhuqi Zhuqi in in 1990, 1990, bedload bedload sediment sediment took took about about 4.1~11.2%4.1~11.2% of of total total sediment sediment when when streamflow streamﬂow discharge discharge was was about 2000–8000 m3/s, and and thethe ratio ratio of of bedload bedload to to suspended suspended load load is is about about 4.3–12.6% 4.3–12.6% with with an an average average of of 7.5%. 7.5%. As As the the averageaverage suspended suspended sediment sediment transport transport was was 6.06 6.06 million million tons tons at at Zhuqi Zhuqi during during 1983–1992 1983–1992 beforebefore the the construction construction of of Shuikou Shuikou Reservoir, Reservoir, the the average average bedload bedload sediment sediment transport transport was was aboutabout 0.455 0.455 ( (≈≈6.066.06 * *7.5%) 7.5%) million million t. t.Xu Xu et et al. al. (2012) (2012) [23] [23 ]estimated estimated that that the the average average annual annual bedload sediment below the Three Gorges Dam during 2003~2010 after the impoundment of the reservoir in 2003 decreased by 93% in comparison with that during 1991–2002. As the Shuikou dam controls 96.2% of the total drainage area above Zhuqi, it is estimated that after the completion of Shuikou reservoir, about 0.41 million tons (≈ 7.5% * 93% * 96.2% * 6.06 million t) of bedload sediment was deposited in the Shuikou reservoir every year due to the reservoir’s retention. Therefore, after the completion of Shuikou reservoir, an average of 3.99 million t of sediment (including 3.58 million t of suspended load and 0.41 million t of bedload) was deposited in the reservoir area every year compared with the 10 years before the completion. Assuming a sediment density of 1300 kg/m3, the Shuikou dam area loses about 3.07 million m3 of its capacity per year due to sediment deposition, which means the reservoir loses about 1.2% of its initial capacity (2.6 billion m3) every 10 years. The double mass curve method is further applied to the investigation of the change point detection for the consistency in the relationship between runoff and sediment load. We plot the double-mass curves of cumulative annual runoff versus cumulative annual SSL at Zhuqi in Figure 6a and cumulative runoff in June versus cumulative sediment load in June in Figure 6b. It is shown that there is a slope break in 1993 in Figure 6a, indicating the annual total amount of sediment significantly changed in 1993, but in June, when the runoff and sediment are the most over the year, significant changes occurred in 1984 (see Figure 6b. The year 1993 is when the largest reservoir, Shuikou dam, started the impoundment, whereas 1984 is the year when the Shaxikou dam, which is the second largest reservoir in the drainage area (25,562 km2) in the Min River Basin, started its cofferdam construction.

Water 2021, 13, 673 10 of 19

bedload sediment below the Three Gorges Dam during 2003~2010 after the impoundment of the reservoir in 2003 decreased by 93% in comparison with that during 1991–2002. As the Shuikou dam controls 96.2% of the total drainage area above Zhuqi, it is estimated that after the completion of Shuikou reservoir, about 0.41 million tons (≈ 7.5% * 93% * 96.2% * 6.06 million t) of bedload sediment was deposited in the Shuikou reservoir every year due to the reservoir’s retention. Therefore, after the completion of Shuikou reservoir, an average of 3.99 million t of sediment (including 3.58 million t of suspended load and 0.41 million t of bedload) was deposited in the reservoir area every year compared with the 10 years before the completion. Assuming a sediment density of 1300 kg/m3, the Shuikou dam area loses about 3.07 million m3 of its capacity per year due to sediment deposition, which means the reservoir loses about 1.2% of its initial capacity (2.6 billion m3) every 10 years. The double mass curve method is further applied to the investigation of the change point detection for the consistency in the relationship between runoff and sediment load. We plot the double-mass curves of cumulative annual runoff versus cumulative annual SSL at Zhuqi in Figure6a and cumulative runoff in June versus cumulative sediment load in June in Figure6b. It is shown that there is a slope break in 1993 in Figure6a, indicating the annual total amount of sediment signiﬁcantly changed in 1993, but in June, when the runoff and sediment are the most over the year, signiﬁcant changes occurred in 1984 (see Figure6b). The year 1993 is when the largest reservoir, Shuikou dam, started the im- Water 2021, 13, x FOR PEER REVIEW poundment, whereas 1984 is the year when the Shaxikou dam, which is the second largest11 of 20 reservoir in the drainage area (25,562 km2) in the Min River Basin, started its cofferdam construction.

Figure 6. The double-mass curve of (a) cumulative annual runoff versus cumulative annual suspended sediment load and Figure(b )6. cumulative The double-mass runoff in curve June versus of (a) cumulativecumulative sediment annual loadrunoff in Juneversus at Zhuqi.cumulative annual Scheme 300. non-overlapping discharge classes between the period 1952–1992 (Period-I) and the period 1994–2018 (Period-II) observed at Zhuqi in the Min River are plotted in Figure 7.The A dailyvisual streamflow inspection andof Figure sediment 7 indicates concentration~streamflow the significant reduction (C~Q) relationshipin sediment concentration for any water dischargewith 300 classes. non-overlapping discharge classes between the period 1952–1992 (Period-I) and the period 1994–2018 (Period-II) observed at Zhuqi in the Min River are plotted in Figure7. A visual inspection of Figure7 indicates the significant reduction in sediment concentration for any water discharge classes.

Figure 7. The sediment rating curves between suspended sediment concentration (C) and streamflow discharge with 300 discharge classes during the period 1952–1992 (Period-I) and the period 1994–2018 (Period-II) observed at Zhuqi in the Min River.

3.3. Changes of Water Level and Riverbed along the Lower Min River Below Shuikou Dam Due to the changes in flow regime, especially in sediment discharge, jointly with the influence of human sand mining activities, the shape of the riverbed in the lower reaches of the Min River is constantly changing. Such kind of changes is especially exhibited in the downcutting in the thalweg of lower Min River. After the Shuikou Reservoir was completed, the thalweg at many river cross-sections along the lower reaches of the Min River got deepened generally and oscillated to varying degrees. As shown in Figure 8, the maximum incision depth in the riverbed during 2000 and 2015 exceeds 20 m at Zhuqi. Figure 9 shows the changes of the thalweg of the lower reaches of the Min River from the Shuikou Dam to the estuary in 2011 and 2015.

Water 2021, 13, x FOR PEER REVIEW 11 of 20

Figure 6. The double-mass curve of (a) cumulative annual runoff versus cumulative annual Scheme 300. non-overlapping discharge classes between the period 1952–1992 (Period-I) and the period 1994–2018 (Period-II) observed at Zhuqi in the Water 2021, 13, 673 11 of 19 Min River are plotted in Figure 7. A visual inspection of Figure 7 indicates the significant reduction in sediment concentration for any water discharge classes.

FigureFigure 7.7. TheThe sedimentsediment ratingrating curves curves between between suspended suspended sediment sediment concentration concentration (C )(C and) and streamﬂow dischargestreamflow with discharge 300 discharge with 300 classes discharge during classe the periods during 1952–1992 the period (Period-I) 1952–1992 and the(Period-I) period and 1994–2018 the (Period-II)period 1994–2018 observed (Period-II) at Zhuqi observed in the Min at River. Zhuqi in the Min River.

3.3.3.3. Changes of Water LevelLevel andand RiverbedRiverbed alongalong thethe LowerLower MinMin RiverRiver belowBelowShuikou Shuikou Dam Dam DueDue toto thethe changeschanges inin ﬂowflow regime,regime, especiallyespecially inin sedimentsediment discharge,discharge, jointlyjointly withwith thethe inﬂuenceinfluence ofof humanhuman sandsand miningmining activities,activities, thethe shapeshape ofof thethe riverbedriverbed inin thethe lowerlower reachesreaches ofof thethe MinMin River River is is constantly constantly changing. changing. Such Such kind kind of changesof changes is especially is especially exhibited exhibited in the in downcuttingthe downcutting in the in thalwegthe thalweg of lower of lower Min Min River. River. AfterAfter thethe ShuikouShuikou ReservoirReservoir waswas completed,completed, thethe thalwegthalweg atat manymany riverriver cross-sectionscross-sections alongalong thethe lowerlower reachesreaches ofof thethe MinMin RiverRiver gotgot deepeneddeepened generallygenerally andand oscillatedoscillated toto varyingvarying degrees. As shown in Figure8, the maximum incision depth in the riverbed during 2000 degrees. As shown in Figure 8, the maximum incision depth in the riverbed during 2000 and 2015 exceeds 20 m at Zhuqi. Figure9 shows the changes of the thalweg of the lower and 2015 exceeds 20 m at Zhuqi. Figure 9 shows the changes of the thalweg of the lower reaches of the Min River from the Shuikou Dam to the estuary in 2011 and 2015. reaches of the Min River from the Shuikou Dam to the estuary in 2011 and 2015.

Figure 8. 8_newRiver bed elevation at the cross-section of the Min River at Zhuqi in different years.

2011 2015 Elevation (m)

(km)

Figure 9. Comparison of thalwegs of the lower Min River (start from the Shuikou Dam to the river mouth) in 2011 and 2015.

Water 2021, 13, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/water

FigureWater 2021 8_new, 13, 673 12 of 19

2011 2015

Elevation (m)

(km)

Figure 9. Comparison of thalwegsFigure 9. of Comparison the lower Min of Riverthalwegs (start of from the lower the Shuikou Min Rive Damr (start to the from river the mouth) Shuikou in 2011 Dam and to 2015.the river mouth) in 2011 and 2015. River channel erosion below dams is a common phenomenon all over the world (ICOLD, 2009) [24]. Similar to the impact of the construction of the Shuikou Reservoir in the Min River, many examples of serious channel incision below dams in relation to regional geographic conditions, e.g., 6 m downcutting below Davis Dam, Colorado River, USA (Mathias, 1997) [25]; 7.5 m downcutting below Hoover Dam, Colorado River, USA (ICOLD, 2009) [24]; 6 m downcutting below Saulspoort Dam, Ash River, South Africa (ICOLD, 2009) [24]; and 10 m downcutting below Three Gorges Dam, Yantze River, China (Zheng et al., 2018) [9]. However, bed erosion is not the only possibility and local conditions Figure 8_newmay determine a variety of outcomes (Lai et al., 2017) [26] in terms of the incision depth or the time to reach a new balance between erosion and deposition. A recent study showed that the new long-term hydro-morphological equilibrium may have been established in the middle and lower Yangtze about 14 years after the closure of TGD (Lai et al., 2017) [26]. In 2011 2015 the case of the Shuikou Dam, such a new equilibrium is still yet to be achieved. Although there is not too much cutting down along the cross-section at Zhuqi during 2008 to 2014 (as shown in Figure7), indicating that the overall elevation of the riverbed got more or less stable at Zhuqi, the thalweg is still getting deeper generally along the lower reach of Min up to 2015 (as shown in Figure8). We can deduce that the riverbed is still in a process of Elevation (m) downward erosion since the Shuikou Dam has been built for more than 20 years. The lowest water level at several locations along the lower reach of the Min River (see Figure 10) shows that the lowest water level of the lower reach has decreased signiﬁcantly (km) since late 1980s’, except at the mouth of the river (Baiyantan). The decline of the water level accelerated in theFigure late 9. Comparison 1990s. of thalwegs of the lower Min River (start from the Shuikou Dam to the river mouth) in 2011 and 2015. Water 2021, 13, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/water

Figure 10. The variation of the lowest water level along the lower reach of the Min River.

Water 2021, 13, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/water Water 2021, 13, x FOR PEER REVIEW 13 of 20

The lowest water level at several locations along the lower reach of the Min River (see Figure 10) shows that the lowest water level of the lower reach has decreased significantly since late 1980s’, except at the mouth of the river (Baiyantan). The decline of the water level accelerated in the late 1990s.

Water 2021, 13, 673 Figure 10. The variation of the lowest water level along the lower reach of the Min River. 13 of 19

The stage-discharge rating curves at Zhuqi station in 1950, 1970, 1990, 2000, 2010, and 2017 areThe plotted stage-discharge in Figure 11. rating A comparison curves at Zhuqi of these station curves in 1950, shows 1970, that 1990, the 2000, water 2010, level and with2017 a specific are plotted discharge in Figure changed 11. A comparisonvery little du ofring these 1950 curves to 1970, shows but thatdropped the water significantly level with 3 aftera speciﬁc late 1990s. discharge With a changed discharge very of little2000 duringm /s, the 1950 average to 1970, water but droppedlevel was signiﬁcantly 8.28, 8.3, 7.73, after 6.93,late 3.73, 1990s. and With 3.60 a m discharge in 1950, of1970, 2000 1990, m3/s, 2000, the average2010, and water 2017, level respectively, was 8.28, 8.3,that 7.73, is, the 6.93, average3.73, and water 3.60 level m in dropped 1950, 1970, 4.68 1990, m in 2000, 68 years 2010, and and such 2017, a respectively,drop mostly that occurred is, the in average the lastwater 30 years. level dropped 4.68 m in 68 years and such a drop mostly occurred in the last 30 years.

2017 2010 2000

Discharge(cms) 1990 1970 1950 100 200 500 2000 5000 20000

2 4 6 8 10 12 14 Water level (m)

FigureFigure 11. 11.TheThe stage-discharge stage-discharge rating rating curves curves at Zhuqi at Zhuqi station station in different in different years. years. 4. Contributing Factors of Sediment Reduction 4. Contributing Factors of Sediment Reduction River sediment comes from many sources including the erosion from land surfaces, River sediment comes from many sources including the erosion from land surfaces, stream banks and stream beds. Although the runoff is the direct driving force of transport stream banks and stream beds. Although the runoff is the direct driving force of transport sediment in the river, the fundamental driving force of the erosion comes from the precipi- sediment in the river, the fundamental driving force of the erosion comes from the tation. Therefore, changes of the amount and intensity of precipitation, land coverage of precipitation.vegetation, andTherefore, the construction changes ofof a the series amount of dams and may intensity result in of changes precipitation, of sediment land flux coveragein the river. of vegetation, and the construction of a series of dams may result in changes of sediment flux in the river. 4.1. Precipitation First of all, we calculated the areal annual precipitation and the annual runoff coefficient in the drainage area above Zhuqi Station, as shown in Figure 12. It can be seen that the annual precipitation of Min River basin increased slightly in the period from 1950 to 2017, while the annual runoff coefficient decreased slightly. However, MK trend test and Pettitt change point test found that these changes were not significant. Further comparison of the correlation between annual precipitation and annual runoff before and after 1993 found that (see Figure 13), after 1993, the slope of the linear regression line between annual runoff and annual precipitation gets smaller, indicating that the annual runoff is less sensitive to the variation of precipitation or, in another word, the variation of annual precipitation results less variation of annual runoff. This implies that a large number of large reservoirs in the Min river basin have the capacity of modulate the variation of annual runoff by reducing the outflow in wet years and increasing the outflow in dry years. The change of precipitation process affects not only runoff process but also sediment yield. The change of precipitation has strong effects on sediment yield. With the increase in rainfall intensity, the erosion capability of rainfall increases, which may lead to the increase in soil loss. According to the USLE soil loss equation (Wischmeier, 1959) [27], one rainfall event is considered to be erosive if the total precipitation amount p ≥ 12.7 mm or the amount of precipitation in 15 min p 15 ≥ 6.4 mm. However, the standards of erosive rainfall vary from region to region. In China, the rainfall amount varies between 9 and 14 mm (Xie et al., 2000) [28]. Huang et al. (2015) [29] investigated the rainfall erosivity in Changting County in Fujian Province and found that the criterion of erosive rainfall was p ≥ 13 mm per day. We investigated the maximum 1-day precipitation and the total amount of erosive precipitation with daily rainfall p ≥ 13 mm from 1950 to 2017 observed at Yangkou in the center of Min River basin and Zhuqi in the lower part of Min River basin. Water 2021, 13, x FOR PEER REVIEW 14 of 20

4.1. Precipitation First of all, we calculated the areal annual precipitation and the annual runoff coefficient in the drainage area above Zhuqi Station, as shown in Figure 12. It can be seen that the annual precipitation of Min River basin increased slightly in the period from 1950 to 2017, while the annual runoff coefficient decreased slightly. However, MK trend test and Pettitt change point test found that these changes were not significant. Further comparison of the correlation between annual precipitation and annual runoff before and Water 2021, 13, 673 after 1993 found that (see Figure 13), after 1993, the slope of the linear regression14 ofline 19 between annual runoff and annual precipitation gets smaller, indicating that the annual runoff is less sensitive to the variation of precipitation or, in another word, the variation of annual precipitation results less variation of annual runoff. This implies that a large AsWater shown 2021, 13 in, x FOR Figure PEER 14 REVIEW, the maximum 1-day precipitation and the total erosive precipitation 2 of 2 number of large reservoirs in the Min river basin have the capacity of modulate the ofvariation Yangkou of andannual Zhuqi runoff showed by reducing no significant the outflow changes, in wet and years MK trend and increasing test and Pettitt the outflow change pointin dry test years. also confirmed no significant changes. That is to say, the change of Min river sediment discharge has veryFigure little 10. to The do variation with the of changethe lowest of water precipitation. level along the lower reach of the Min River.

Figure 12.Figure FigureVariation 12. 12_new Variationof (a) annual of ( aprecipitation) annual precipitation and (b) runoff and ( bcoefficient) runoff coefﬁcient above Zhuqi. above Zhuqi.

Water 2021, 13, x FOR PEER REVIEW 15 of 20

amount of erosive precipitation with daily rainfall p ≥ 13 mm from 1950 to 2017 observed at Yangkou in the center of Min River basin and Zhuqi in the lower part of Min River basin. As shown in Figure 14, the maximum 1-day precipitation and the total erosive precipitation of Yangkou and Zhuqi showed no significant changes, and MK trend test

and Pettitt change point test also confirmed no significant changes. That is to say, the Figure 15_b_newchange of Min river sediment discharge has very little to do with the change of FigureFigure 13. TheThe correlation correlation between between annual annual areal areal precip precipitationitation and and annual annual runoff runoff at atZhuqi Zhuqi in the in the precipitation. periodperiod beforebefore andand afterafter 1993.1993.

The change of precipitation process affects not only runoff process but also sediment yield. The change of precipitation has strong effects on sediment yield. With the increase in rainfall intensity, the erosion capability of rainfall increases, which may lead to the increase in soil loss. According to the USLE soil loss equation (Wischmeier, 1959) [27], one rainfall event is considered to be erosive if the total precipitation amount p ≥ 12.7 mm or the amount of precipitation in 15 min p 15 ≥ 6.4 mm. However, the standards of erosive rainfall vary from region to region. In China, the rainfall amount varies between 9 and 14 mm (Xie et al., 2000) [28]. Huang et al. (2015) [29] investigated the rainfall erosivity in Changting County in Fujian Province and found that the criterion of erosive rainfall was p ≥ 13 mm per day. We investigated the maximum 1-day precipitation and the total

Figure 14. Variation of the maximum 1-day precipitation and the total erosive precipitation of Figure 14. VariationYangkou of andthe maximum Zhuqi. (a 1-day) Maximum precipitation 1-day and precipitation the total erosive of Yangkou, precipitation (b) total of Yangkou erosive precipitationand Zhuqi. (a) of Maximum 1-dayYangkou, precipitation (c) maximum of Yangkou, 1-day (b) total precipitation erosive precipitation of Zhuqi, andof Yangkou, (d) total (c erosive) maximum precipitation 1-day precipitation of Zhuqi. of Zhuqi, and (d) total erosive precipitation of Zhuqi.

4.2. Vegetation The changes of sediment volume and sediment grain size is not only affected by the operation of reservoirs but the increase in vegetation cover over the basin may also play an important role in the variation of sediment generation process in the Min River basin. Using NASA’s GIMMS (Global Inventory Modeling and Mapping Studies) NDVI (normalized difference vegetation index) data product during 1981–2015 (https://glam1.gsfc.nasa.gov), we calculated the annual areal average NDVI over the Min River basin, plotted in Figure 15a, which shows an obvious upward trend. Mann-Kendall trend test is also conducted for monthly NDVI on a grid basis, and the result is presented in Figure 15b. It is clearly shown that the increase in NDVI is very significant over most parts of the Min River basin. Such kind of significant increase in vegetation cover inevitably could lead to the decrease in sediment yield because vegetation restoration can effectively prevent soil loss (Wang et al., 2016) [30].

Water 2021, 13, 673 15 of 19

4.2. Vegetation The changes of sediment volume and sediment grain size is not only affected by the operation of reservoirs but the increase in vegetation cover over the basin may also play an important role in the variation of sediment generation process in the Min River basin. Using NASA’s GIMMS (Global Inventory Modeling and Mapping Studies) NDVI (normalized difference vegetation index) data product during 1981–2015 (https://glam1.gsfc.nasa. gov), we calculated the annual areal average NDVI over the Min River basin, plotted in Figure 15a, which shows an obvious upward trend. Mann-Kendall trend test is also conducted for monthly NDVI on a grid basis, and the result is presented in Figure 15b. It is clearly shown that the increase in NDVI is very signiﬁcant over most parts of the Min Water 2021, 13, x FOR PEER REVIEW River basin. Such kind of signiﬁcant increase in vegetation cover inevitably could lead16 of to 20

Water 2021, 13, x FOR PEER REVIEWthe decrease in sediment yield because vegetation restoration can effectively prevent16 of soil20

loss (Wang et al., 2016) [30].

(a) (b(b) )

FigureFigure 15.15. ChangesChanges ofof thethe annualannual arealareal average averageaverage NDVI NDVINDVI (( a(a)a) and)and and the the the trend trend trend of of of monthly monthly monthly NDVI NDVI NDVI ( b( b)(b) over over) over the the Min River basin during 1981 to 2015. (Note: the number +3, +2, +1, and 0 stand for very Minthe Min River River basin basin during during 1981 1981 to 2015. to 2015. (Note: (Note: the the number number +3, +3, +2, +2, +1, +1, and and 0 0 stand stand for for very very signiﬁcant significantsignificant trend,trend, significant trend, weak trend,trend, andand nono trend, trend, respectively, respectively, and and the the sign sign (+ (+ or or − )− ) trend, signiﬁcant trend, weak trend, and no trend, respectively, and the sign (+ or −) stands for standsstands forfor negativenegative or positive trend.). negative or positive trend.).

4.3.4.3. Dam Dam Construction Construction TheThe Min Min river is rich in runoff, but but the the runoff runoffrunoff varied variedvaried seasonally seasonally due due the the effects effects of of monsoonmonsoon climate.climate. To To adjust adjust the the seasonalseasonal distributiondistribution ofof runoff, runoff, and and meanwhile meanwhilemeanwhile take taketake advantageadvantage ofof the the rich rich hydropower hydropower in the inin river ththe basin,e riverriver numerous basin,basin, numerousnumerous reservoirs andreservoirsreservoirs hydropower andand plantshydropowerhydropower have been plants constructed, have been including constructed, nine inin largecludingcluding reservoirs ninenine large large with reservoirs capacityreservoirs over with with 100 capacity capacity million overmover3 as 100100 shown millionmillion in Figurem3 as shown1. Figure inin FigureFigure16 shows 1.1. FigureFigure the signiﬁcant 1616 showsshows increasethe the significant significant in the increase totalincrease reservoir in in the the totalcapacitytotal reservoirreservoir of the capacitycapacity reservoirs of inthe the reservoirsreservoirs Min River inin thebasinthe MinMin during River River 1954 basin basin to du du 2011.ringring 1954 1954 to to 2011. 2011. ) 3 ) 3 m 6 m 6 Total capacity ( 10 ( capacity Total Total capacity ( 10 ( capacity Total 0 2000 4000 6000 8000 0 2000 4000 6000 8000

1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010 Year Year Figure 16. The change of total reservoir capacity in the Min River basin during 1954–2011. Figure 16. The change of total reservoir capacity in the Min River basin during 1954–2011. The effects of dams on sediment generation and transport are mixed. After the The effects of dams on sediment generation and transport are mixed. After the impoundment of a reservoir, more sediment will deposit in the reservoir area, resulting impoundment of a reservoir, more sediment will deposit in the reservoir area, resulting in the reduction in sediment transport to the downstream. At the same time, due to the in the reduction in sediment transport to the downstream. At the same time, due to the decrease in sediment concentration in the streamflow, the erosivity of streamflow will decrease in sediment concentration in the streamflow, the erosivity of streamflow will increase, which leads to more intensive erosion of the downstream river channel. increase, which leads to more intensive erosion of the downstream river channel. However, in general, the sediment reduction caused by the sedimentation in reservoirs is However, in general, the sediment reduction caused by the sedimentation in reservoirs is much greater than the sediment increase caused by downstream channel erosion. For much greater than the sediment increase caused by downstream channel erosion. For example, the curve of cumulative annual runoff versus cumulative annual sediment example,transport theat Zhuqicurve inof Figurecumulative 6 shows annual that runoff the abrupt versus decrease cumulative in annual annual sediment sediment transporttransport inat 1993Zhuqi and in the Figure abrupt 6 decreaseshows that in sedimentthe abrupt transport decrease during in annual the main sediment flood transportseason in in1984 1993 are and respectively the abrupt relateddecrease to inthe sediment construction transport of Shuikou during thereservoir main floodand seasonShaxikou in reservoir,1984 are whichrespectively are the related first and to the the second construction in drainage of Shuikou area in the reservoir Min River and Shaxikoubasin. On reservoir,the other hand,which the are construction the first and of the other second major in reservoirs drainage abovearea in Zhuqi the Min have River no basin.significant On the impacts other onhand, the theSSL construction at Zhuqi. That of means,other major the significant reservoirs sediment above Zhuqi reduction have isno mostly caused by those large reservoirs, which take large portions of the drainage area.

Water 2021, 13, 673 16 of 19

The effects of dams on sediment generation and transport are mixed. After the impoundment of a reservoir, more sediment will deposit in the reservoir area, resulting in the reduction in sediment transport to the downstream. At the same time, due to the decrease in sediment concentration in the streamflow, the erosivity of streamflow will increase, which leads to more intensive erosion of the downstream river channel. However, in general, the sediment reduction caused by the sedimentation in reservoirs is much greater than the sediment increase caused by downstream channel erosion. For example, the curve of cumulative annual runoff versus cumulative annual sediment transport at Zhuqi in Figure6 shows that the abrupt decrease in annual sediment transport in 1993 and the abrupt decrease in sediment transport during the main flood season in 1984 are respectively related to the construction of Shuikou reservoir and Shaxikou reservoir, which are the first and the second in drainage area in the Min River basin. On the other hand, the construction of other major reservoirs above Zhuqi have no significant impacts on the SSL at Zhuqi. That means, the significant sediment reduction is mostly caused by those large reservoirs, which take large portions of the drainage area.

4.4. Sand Mining Sand mining is another major cause of sediment reduction. There is a huge demand for river sand, and large-scale sand mining occurred in the lower reaches of the Min River. There is no accurate data about the amount of river sand mining because many sand mining operations are illegal. The annual river sand consumption in the lower Min is estimated to be more than 10 million tons, which is much more than the sediment transported to the lower Min from the upstream reaches in recent years after the construction of Shuikou dam in 1993. When sand mining removes huge amount of sediment, signiﬁcant river bed incision is inevitable (e.g., Hackney et al., 2020) [31]. Figure8 shows the dramatic deepening of the channel by over 10 m between 2003 and 2008 at Zhuqi, whereas the bathymetric changes between 2000–2003 and 2008–2014 are not so signiﬁcant compared to the change between 2003 and 2008. Moreover, the annual average SSC and annual SSL on Figure5 show no indication of huge reduction between 2003 and 2008. Therefore, we can deduce that sand removal by mining activities plays a very important role in the deepening of the lower Min River channel, especially between 2003 and 2008.

4.5. Relative Attribution of Sediment Discharge Changes The change of areal annual precipitation observed in the drainage basin above Zhuqi and annual average streamflow discharge, average suspended sediment concentration, and suspended sediment load observed at Zhuqi during period-1 (before 1993) and period-2 (after 1993) are summarized in Table3. Here, E0 is estimated using the meteorological variables observed at 19 sites within the Min River basin with the FAO-56 method (1976); the catchment property (m) is estimated using the Equations (10) and (11). According to the statistics presented in Table3, we know that the changes of precipitation, E0, and m between period-1 and period-2 are very small (3%, −0.5%, and −3.4%, respectively), whereas the change of sediment concentration (C) is considerable (−70.5%). For a given catchment, many of the catchment properties (such as topography, soil type/depth, geologic substrate, etc.) remain nearly constant except for the land cover. Therefore, the change in the value of m mostly reflect the change of land cover, and a change in vegetation cover from grasses to trees would increase m (Roderick and Farquhar, 2011) [17]. However, the estimated m for the Min River basin decreased in the period 2, which seems to be in contradiction to the fact that the average NDVI significantly increased during 1981~2015 (as shown in Figure 15). Such a contradiction likely resulted from the uncertainty in the estimation of areal precipitation and E0. Water 2021, 13, 673 17 of 19

Table 3. Changes of areal annual precipitation (P), potential evaporation (E0), and catchment property (m) in the drainage area above Zhuqi, and annual average streamﬂow discharge (Q), average suspended sediment yield (QS), and suspended sediment concentration (C) observed at Zhuqi during period-1 (1951–1992) and period-2 (1993–2017).

2 3 P (mm/year) E0 (mm/year) Q (mm/year) QS (t/km /year) C (kg/m ) m (−)

P1 ∆PE0,1 ∆E0 Q1 ∆QQS,1 ∆QS C1 ∆C m1 ∆m 1653.7 50.6 1087.9 −5.4 973.5 +60.3 131.2 −85.7 0.129 −0.091 1.056 −0.037 Note: the subscript “1” indicates the value in period 1; ∆ is the value in period 2 minus the value in period 1.

The elasticity indices ηQ and ηC for the Min River are computed using Equations (7) and (8) for different number of discharge classes (n), and the results are plotted in Figure 17. For n Water 2021, 13, x FOR PEER REVIEW 18 of 20 approximately greater than 800, the variations of ηQ and ηC are generally stable, and we take ηQ = 0.95 and ηC = 1.09 at n = 1000 as the representative values of ηQ and ηC. Elasticity

ηC ηQ 0.20.40.60.81.0

0 500 1000 1500 2000 2500 3000 n

FigureFigure 17. 17. VariationVariation of of elasticity elasticity indices indices ηηQQ andand ηηC Cforfor the the Min Min River River vary vary with with increasing increasing nn. .

ThenThen we we can can calculate calculate the the elasticity elasticity of of suspended suspended sediment sediment load load (Qs (Qs) to) toP, PE,0E and0 and m withm with Equation Equation (12), (12), and and the the relative relative contribution contribution of of different different contributing contributing factors factors with with EquationEquation (14). TheThe resultsresults are are presented presented in Tablein Ta4ble. The 4. absoluteThe absolute values values of the elasticof the indiceselastic in Table4 showed that ηP >ηC >ηm > ηE0, indicating that the sensitivity of Qs to these indices in Table 4 showed that 𝜂 >𝜂 >𝜂 > 𝜂, indicating that the sensitivity of Qs to thesevariables variables can be can ranked be ranked as precipitation as precipitation > sediment > sediment concentration concentration (C) > catchment (C) > catchment property property(m) > potential (m) > potential evapotranspiration evapotranspiration (E0). Although (E0). CAlthoughis not the C most is not inﬂuential the most factorinfluential to Qs factorchange, to itQs changed change,the it changed most between the most period-1 between and period-1 period-2, and whereas period-2, there whereas is little there change is in precipitation, E0, and m, hence the contribution to Qs reduction is dominated by the little change in precipitation, E0, and m, hence the contribution to Qs reduction is change of C. The sum of εs,m and εs,C, which represents the contribution of human activities dominated by the change of C. The sum of 𝜀, and 𝜀, , which represents the (including land use/cover change and dam construction), is estimated to contribute 103.8% contribution of human activities (including land use/cover change and dam construction), of the reduction in suspended sediment, whereas climate variability only contributed is estimated to contribute 103.8% of the reduction in suspended sediment, whereas climate −3.8% to the reduction in suspended sediment. variability only contributed −3.8% to the reduction in suspended sediment.

Table 4. Elasticity of suspended sediment load (Qs) and contribution to Qs reduction with respect Table 4. Elasticity of suspended sediment load (Qs) and contribution to Qs reduction with respect toto precipitation precipitation (P (P),), potential potential evapotranspiration evapotranspiration (E (E0),0), sediment sediment concentration concentration (C (),C ),and and catchment catchment propertyproperty ( (mm).).

ElasticityElasticity of of QQs s ContributionContribution to Qss ReductionReduction (%) (%)

η η η ηm η ε ε εs,m ε 𝜼𝑸Q 𝜼𝑷P 𝜼𝑬𝟎E0 𝜼𝒎 𝜼𝑪C 𝜺s,Ps,P 𝜺s,Es,E0 0 𝜺s,m 𝜺s,C 0.95 1.34 −0.39 −0.41 1.09 −3.6 −0.2 −1.3 105.1 0.95 1.34 −0.39 −0.41 1.09 −3.6 −0.2 −1.3 105.1

5. Conclusions Although there was no significant change of precipitation during 1950–2017, the construction of a series of dams along the upper reaches of the Min River, especially the Shuikou Dam, which started filling in 1993, modified the flow processes, leading to the significant increase in low-flows and slightly decrease in flood-flows at Zhuqi located at the lower Min River. At the same time, the variability of annual runoff is less sensitive to the variability of precipitation because large reservoirs can modulate the variation of annual runoff by reducing the outflow in wet years and increasing the outflow in dry years. Reservoirs have more effects on the sediment transport process than flow process by trapping most sediment in the reservoirs, and greatly reduced the amount of sediment transporting downstream. Vegetation over the Min River basin increased significantly in the last four decades, and the increase in vegetation cover also contributes to the decrease in sediment yield. The reduction in sediment together with excessive sand mining in the lower Min River resulted in the severe downcutting of the riverbed. The average water

Water 2021, 13, 673 18 of 19

5. Conclusions Although there was no significant change of precipitation during 1950–2017, the construction of a series of dams along the upper reaches of the Min River, especially the Shuikou Dam, which started filling in 1993, modified the flow processes, leading to the significant increase in low-flows and slightly decrease in flood-flows at Zhuqi located at the lower Min River. At the same time, the variability of annual runoff is less sensitive to the variability of precipitation because large reservoirs can modulate the variation of annual runoff by reducing the outflow in wet years and increasing the outflow in dry years. Reservoirs have more effects on the sediment transport process than flow process by trapping most sediment in the reservoirs, and greatly reduced the amount of sediment transporting downstream. Vegetation over the Min River basin increased significantly in the last four decades, and the increase in vegetation cover also contributes to the decrease in sediment yield. The reduction in sediment together with excessive sand mining in the lower Min River resulted in the severe downcutting of the riverbed. The average water level observed at Zhuqi dropped 4.68 m during 1950 to 2017, and such a drop mostly occurred after the construction of Shuikou Dam. Using a reformulated elasticity approach to quantifying climatic and anthropogenic contributions to sediment changes, the relative contribution of precipitation variability and human activities to sediment reduction in the lower Min River are quantified, which shows that the sediment reduction is fully caused by human activities (including land use/land cover changes and dam construction).

Author Contributions: Conceptualization and methodology, W.W.; calculation and analysis, T.W. and W.C.; data collection and processing, B.C., Y.Y., and J.W.; project administration, F.M.; writing, all authors. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by National Science Foundation of China project (grant number 419710420) and Water Conservancy Science and Technology Project of Fujian Province. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References 1. Magilligan, F.J.; Nislow, K.H. Changes in hydrologic regime by dams. Geomorphology 2005, 71, 61–78. [CrossRef] 2. Wang, W.; Wang, X.-G.; Zhou, X. Impacts of Californian dams on flow regime and maximum/minimum flow probability distribution. Hydrol. Res. 2011, 42, 275–289. [CrossRef] 3. Nilsson, C.; Reidy, C.A.; Dynesius, M.; Revenga, C. Fragmentation and flow regulation of the world’s large river systems. Science 2005, 308, 405–408. [CrossRef] 4. Gupta, H.; Kao, S.J.; Dai, M. The role of mega dams in reducing sediment fluxes: A case study of large Asian rivers. J. Hydrol. 2012, 464–465, 447–458. [CrossRef] 5. Yang, S.L.; Xu, K.H.; Milliman, J.D.; Yang, H.F.; Wu, C.S. Decline of Yangtze River water and sediment discharge: Impact from natural and anthropogenic changes. Sci. Rep. 2015, 5, 12581. [CrossRef][PubMed] 6. Binh, D.V.; Kantoush, S.A.; Sumi, T. Changes to long-term discharge and sediment loads in the Vietnamese Mekong Delta caused by upstream dams. Geomorphology 2020, 353, 107011. [CrossRef] 7. Song, C.; Wang, G.; Sun, X.; Chang, R.; Mao, T. Control factors and scale analysis of annual river water, sediments and carbon transport in China. Sci. Rep. 2016, 6, 25963. [CrossRef] 8. Zhang, S.-H.; Li, Z.-H.; Hou, X.-N.; Yi, Y.-J. Impacts on watershed-scale runoff and sediment yield resulting from synergetic changes in climate and vegetation. Catena 2019, 179, 129–138. [CrossRef] 9. Zheng, S.-W.; Xu, Y.-J.; Cheng, H.; Wang, B.; Xu, W.; Wu, S. Riverbed erosion of the final 565 km of the Yangtze River (Changjiang) following construction of the Three Gorges Dam. Sci. Rep. 2018, 8, 11917. [CrossRef] 10. Xia, J.; Deng, S.; Lu, J.; Xu, Q.; Zong, Q.; Tan, G. Dynamic channel adjustments in the Jingjiang Reach of the Middle Yangtze River. Sci. Rep. 2016, 6, 22802. [CrossRef] 11. Eslami, S.; Hoekstra, P.; Trung, N.N.; Kantoush, S.A.; Vegt, M.V.D. Tidal amplification and salt intrusion in the mekong delta driven by anthropogenic sediment starvation. Sci. Rep. 2019, 9, 18746. [CrossRef][PubMed] 12. Luo, X.-X.; Yang, S.-L.; Wang, R.-S.; Zhang, C.-Y.; Li, P. New evidence of Yangtze delta recession after closing of the Three Gorges Dam. Sci. Rep. 2017, 7, 41735. [CrossRef][PubMed] Water 2021, 13, 673 19 of 19

13. Liu, X.; Paul, J.; Saito, Y.; Nguyen, V.L. Recent evolution of the Mekong Delta and the impacts of dams. Earth-Sci. Rev. 2017, 175, 1–17. [CrossRef] 14. Timpe, T.; Kaplan, D. The changing hydrology of a dammed Amazon. Sci. Adv. 2017, 3, e1700611. [CrossRef] 15. Richter, B.D.; Baumgartner, V.J.; Powell, J.; Braun, D.P. A method for assessing hydrologic alteration within ecosystems. Conserv. Biol. 1996, 10, 1163–1174. [CrossRef] 16. Searcy, J.K.; Hardisoni, C.H.; Langbein, W.B. Double Mass Curves, Geological Survey Water Supply Paper 1541-B; U.S. Geological Survey: Washington, DC, USA, 1960. 17. Kendall, M.G. Rank Correlation Methods, 4th ed.; Charles Griffin: London, UK, 1975. 18. Pettitt, A.N. A Non-Parametric Approach to the Change-Point Problem. J. R. Stat. Soc. 1979, 28, 126–135. [CrossRef] 19. Zhang, J.; Gao, G.; Fu, B.; Gupta, H.V. Formulating an elasticity approach to quantify the effects of climate variability and ecological restoration on sediment discharge change in the Loess Plateau, China. Water Resour. Res. 2019, 55, 9604–9622. [CrossRef] 20. Michael, L.; Farquhar, G.D. A simple framework for relating variations in runoff to variations in climatic conditions and catchment properties. Water Resour. Res. 2011, 47, W00G07. 21. Choudhury, B. Evaluation of an empirical equation for annual evaporation using field observations and results from a biophysical model. J. Hydrol. 1999, 216, 99–110. [CrossRef] 22. Allen, R.G.; Pereira, L.S.; Raes, D.; Smith, M. Crop Evapotranspiration: Guidelines for Computing Crop Water Requirements; FAO Irrigation and Drainage Paper 56; FAO—Food and Agriculture Organization of the United Nations: Rome, Italy, 1998. 23. Xu, Q.-X.; Tong, H. Study on the variation of runoff and sediment in the Yangtze River in recent 50 years. J. China Hydrol. 2012, 32, 38–47. (In Chinese) 24. International Committee on Large Dams. Sedimentation and Sustainable Use of Reservoirs and River Systems; Sedimentation Committee Draft Bulletin; International Committee on Large Dams: Paris, France, 2009. 25. Kondolf, G.M. Profile: Hungry water: Effects of dams and gravel mining on river channels. Environ. Manag. 1997, 21, 533–551. [CrossRef] 26. Lai, X.-D.; Yin, D.-W.; Finlayson, B.L.; Wei, T.-Y.; Li, M.-T.; Yuan, W.; Yang, S.-L.; Dai, Z.-J.; Gao, S.; Chen, Z.-Y. Will river erosion below the Three Gorges Dam stop in the middle Yangtze? J. Hydrol. 2017, 554, 24–31. [CrossRef] 27. Wischmeier, W.H. A rainfall erosion index for a Universal Soil-Loss Equation. Soil Sci. Soc. Am. J. 1959, 23, 246–249. [CrossRef] 28. Xie, Y.; Liu, B.-Y.; Zhang, W.-B. Study on the standard of erosive rainfall. J. Soil Water Conserv. 2000, 14, 6–11. (In Chinese) 29. Huang, L.-P.; Mao, Z.-Y.; Fu, S.-L.; Xu, P.-P. Research on rainfall erosivity and its relationship with soil & water loss in Changting County, Fujian Province. J. Nat. Dis. 2015, 24, 103–111. (In Chinese) 30. Wang, S.; Fu, B.J.; Piao, S.L.; Lü, Y.H.; Ciais, P.; Feng, X.M.; Wang, Y.F. Reduced sediment transport in the Yellow River due to anthropogenic changes. Nat. Geosci. 2016, 9, 38–41. [CrossRef] 31. Hackney, C.R.; Darby, S.E.; Parsons, D.R.; Leyland, J.; Houseago, R.C. River bank instability from unsustainable sand mining in the lower Mekong River. Nat. Sustain. 2020, 3, 217–225. [CrossRef]