Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Catena 188 (2020) 104447

Contents lists available at ScienceDirect

Catena
journal homepage: www.elsevier.com/locate/catena

Effect of gravel content on the sediment transport capacity of overland flow T


Zhenzhi Zhan, Fangshi Jiang, Peisong Chen, Pengyu Gao, Jinshi Lin, Hongli Ge,

Ming Kuang Wang, Yanhe Huang
College of Resource and Environment, Fujian Agriculture and Forestry University, Fuzhou 350002, China

A R T I C LE I N FO A B S T R A C T

Keywords: The sediment transport capacity (Tc) is an important parameter for analysing and controlling soil erosion pro-
Benggang cesses. However, few studies have investigated the Tc of soils containing gravel (grain sizes of > 2 mm) on steep
Colluvial deposit slopes. Colluvial deposits on steep slopes in benggang are composed of loose materials with large amounts of
Gravel content gravel. This study aimed to investigate the effects of gravel content on the Tc of overland flow using colluvial
Overland flow
deposits with a range of slope gradients and flow discharges in a nonerodible flume (4 m long, 0.12 m wide and
Sediment transport capacity
0.1 m high). The experiments were carried out using six gravel mass contents (0, 10, 20, 30, 40, and 50%), four
slope gradients (18, 36, 57, and 84%), and four flow discharges (0. 56, 1.11, 2.22, and 4.44 × 10-3 m2 s−1). The
experimental results revealed that the measured Tc increased linearly as the gravel content increased. The effects
of the slope gradient, flow discharge and gravel content on Tc were significant (p < 0.01). Tc increased fol-
lowing power functions either with slope gradient, flow discharge and gravel content (coefficient of model
determination (r2) = 0.96 and Nash-Sutcliffe model efficiency (NSE) = 0.96), or with slope gradient, flow
discharge and sediment median grain size (d50) (r2 = 0.98 and NSE = 0.98). The two established power
functions facilitated the prediction of the Tc of colluvial deposits in this study. However, the power functions
overestimated the predicted Tc values when the measured Tc was less than 1 kg m−1 s−1 and underestimated the
values when the measured Tc ranged from 2 to 4 kg m−1 s−1. These findings are helpful for improving the
simulation of erosion processes in colluvial deposits and deepening the understanding of erosion mechanisms in
soil and rock mixtures. Moreover, the relationship of hydrodynamic variables with Tc on colluvial deposits
remains to be investigated in additional in-depth studies.

1. Introduction Tc is greatly influenced by the slope gradient, flow discharge and


sediment size, among other conditions. Some general conclusions have
Soil erosion is an important environmental problem and a popular been made, including that Tc increases as the slope gradient and flow
research topic worldwide. Soil erosion involves three processes: soil discharge increase (Abrahams et al., 2001; Ali et al., 2012; Beasley and
particle detachment, transport and deposition (Nearing et al., 1989). Huggins, 1982; Govers, 1990; Jiang et al., 2018a). In most cases, low-
The sediment transport capacity (Tc) of overland flow is a pivotal angle slopes (< 10%) are used to establish the Tc equations used in soil
parameter for analysing and controlling the soil erosion process and erosion models. However, hydraulic mechanisms differ between gentle
plays a vital role in determining the soil detachment rate and sediment and steep slopes. The Tc equations used for gentle slopes are not ne-
transport (Nearing et al., 1997; Lei et al., 2001; Aksoy et al., 2016). Tc is cessarily appropriate for steep slopes. Recent studies have focused on
defined as the maximum load of sediment that a given flow discharge the Tc values of slope gradients greater than 20%. The slope gradient
can carry and is fundamental for developing process-based erosion explored for sand by Zhang et al. (2009) ranged from 9 to 47% and that
models (Nearing et al., 1989). Widely used modelling methods assume explored for loess soils by Wu et al (2016) ranged from 10 to 39%. Jiang
that soil detachment occurs only when the sediment load is below the Tc et al. (2018a) investigated slope gradients varying from 18% to 84% in
limit and that sediment deposition occurs only when the sediment load their study on colluvial soils. Major differences were observed between
exceeds the Tc limit. Based on these assumptions, Tc is extensively ap- gentle and steep slopes in these studies. For example, in Jiang et al.,
plicable in process-based erosion prediction models (Finkner et al., 2018a, the increase in Tc (with slope and flow discharge) observed for
1989; De Roo, 1996; Zhang et al., 2009; Mahmoodabadi et al., 2014). steeper slopes was less than that for shallower slopes. Existing studies


Corresponding author.
E-mail address: yanhehuang@163.com (Y. Huang).

https://doi.org/10.1016/j.catena.2019.104447
Received 29 July 2019; Received in revised form 23 December 2019; Accepted 24 December 2019
0341-8162/ © 2019 Elsevier B.V. All rights reserved.
Z. Zhan, et al. Catena 188 (2020) 104447

have shown that sediment size is another important factor affecting Tc deposits have the important feature of a high gravel mass content (more
and have elucidated the effects of sediment size in some models of Tc. than 40%) (Jiang et al., 2014), but how gravel affects the Tc of overland
For example, Agarwal and Dickinson (1991) and Zhang et al. (2011) flow on colluvial deposits is still unclear and requires further study.
reported that sediment size was negatively correlated with Tc for dif- Many studies on Tc have focused on fine soil particles (particle
ferent sediment median grain sizes. Conversely, the Tc showed a posi- sizes < 2 mm) from different soil materials over the past several
tive correlation with particle size, and it increased with increasing soil decades and ignored the effect of gravel (particle sizes > 2 mm or
mean weight diameter (Low, 1989; Mahmoodabadi et al., 2014). larger) on Tc, and this lack of information restricts the understanding of
Nevertheless, most studies of Tc have focused on fine soil particles and soil erosion processes on colluvial deposits and erosion mechanisms in
have removed gravel (mineral particles 2 mm or larger in diameter) soil and rock mixtures. Furthermore, prediction models based on gravel
from different soil materials over the past several decades (Abrahams content for Tc are not established, which restricts the application of
et al., 2001; Ali et al., 2012; Mu et al., 2019; Nearing et al., 1997; Zhang prediction equations of Tc. Therefore, experimental data on Tc in col-
et al., 2009). Consequently, few studies have considered the effects of luvial deposits with steep slopes containing large volumes of gravel are
gravel (> 2 mm) on Tc. necessary. We conducted a detailed and systematic study to analyse the
Gravel, which can alter soil structure and soil physical character- Tc of different gravel contents in different slope gradients and different
istics, is an important soil factor that cannot be ignored in soil erosion flow discharges, and explore the relationship of Tc with gravel content
processes (Poesen et al., 1994; Cerdà, 2001). Miller and Guthrie (1984) or sediment median grain size. In this study, experiments simulating Tc
defined gravel or rock fragments as particles with a diameter of more with a nonerodible bed were conducted to meet the following objec-
than 2 mm, including all sizes that have horizontal dimensions less than tives: (i) to investigate the effects of slope gradient, flow discharge and
the size of a pedon. Gravel is often a component of soils, and due to its gravel content on the Tc in hydraulic flumes on steep colluvial deposits,
effects on infiltration, interrill runoff, soil hydrological processes and (ii) to quantitatively analyse the changes in Tc based on gravel content
hydraulic properties, it cannot be neglected when considering soil for a range of slope gradients and flow discharges, and (iii) to evaluate
erosion processes (Brakensiek et al., 1994; Figueiredo et al., 1998; the relationships of Tc with slope gradient, flow discharge and gravel
Wang et al., 2015). Hence, gravel is another important factor affecting content on colluvial deposits from collapsing gullies. The hypothesis of
the Tc. Compared with studies investigating the hydraulic conditions of this study was that gravel content would impact the Tc of overland flow
slopes, those quantifying the Tc values of soil materials containing on colluvial deposits. We hope that this study will enrich our under-
gravel on steep slopes are rare. The effects of gravel content on Tc have standing of the mechanisms of erosion processes on colluvial deposits in
not been reported or evaluated under different hydraulic conditions, benggang.
and this lack of information restricts the understanding of soil erosion
and sediment yield processes.
2. Material and methods
Gullies are extremely common in southern China and are referred to
locally as benggang or collapsing gullies (Jiang et al., 2014). A collap-
2.1. Study area
sing gully is formed on hill slopes covered by a thick weathered granite
mantle and is the result of the combined action of hydraulic scouring
The study area is located in Yangkeng village (24°57′ N, 118°03′ E),
and gravitational collapse (Xu, 1996). These gullies develop rapidly,
Longmen town, Anxi County, Fujian Province, southeastern China
and their walls collapse suddenly. Zhong et al. (2013) reported that the
(Fig. 2). This region features a temperate climate throughout the year,
annual average erosion by collapse in these areas was more than 50 kt
with a mean annual temperature of 19 °C, and there are 293–340 frost-
km−2 yr−1. In addition, the hyperconcentrated flow produced by col-
free days per year in this subtropical monsoon region. The mean annual
lapsing gullies can destroy water conservancy structures, roads and
precipitation is 1800 mm, most of which falls from May to September in
houses and can cause serious losses of both life and property (Lin et al.,
the form of thundershowers and rainstorms during the typhoon season
2017). A typical collapsing gully generally consists of four parts, i.e., an
(Lin et al., 2017). Red soils constitute the main soil type in the area and
upper catchment, a collapsing wall, a colluvial deposit, and an alluvial
are classified as mesic Typic Kandiudoxes (Soil Survey Staff, 2014);
fan (Fig. 1). Colluvial deposits (Fig. 1B), where residual material is
these soils are underlain by granitoid rocks and metamorphosed
deposited, become filled with material from the upper catchment and
equivalents. The main rock types in Longmen town comprise pyroxene
collapsing wall because of the effects of water and gravity. Colluvial
diorite, quartz monzonite, and dacite (Chen et al., 2018). These natural
soils have the characteristics of a weak structure, low cohesion, poor
conditions have caused serious soil erosion in the study area.
stability, and high erodibility. The loose soil in these deposits can be
scoured very quickly during rainfall, and this loose soil constitutes the
main source of eroded sediment in collapsing gully systems (Gong et al., 2.2. Experimental soil samples
2011). Dong et al. (2016) noted that the erosion of colluvial deposits
represents approximately 60–80% of the total soil erosion in collapsing The soil materials for the experiments were collected from typical
gullies. The colluvial deposits disintegrate rapidly, and the particle sizes colluvial deposits in Yangkeng village. The soil samples from colluvial
are ideal for water transport after immersion in water. In addition, deposits were air-dried. After being air-dried, equal amounts of five soil
rainfall splash erosion and runoff scouring occur easily because of the samples from these deposits were blended, and the mixture was then
steep slope (slope angles can reach 84%) and lack of plant roots and crushed and passed through a 10 mm sieve. The gravel particles (> 2
organic matter (Jiang et al., 2014). Consequently, colluvial deposits mm) were separated from the fine soil particles (< 2 mm) by sieving.
form more easily from a hyperconcentrated flow under rainfall. To- The soil pH was measured at a soil:water ratio of 1:5 with a Starter
gether, these conditions can directly affect the sediment transport of 2100 pH meter (Ohaus Instruments Co., Ltd., Shanghai, China). The
colluvial deposits. Therefore, the erosion of colluvial deposits requires organic matter content, cation-exchange capacity (CEC) and particle
attention. Previous studies have examined the characteristics of runoff size fraction analyses of the tested soils were performed by standard
and sediment loss on steep colluvial deposits (Jiang et al., 2014; Lin analytical methods in accordance with those of Jackson (1979). The soil
et al., 2017; Jiang et al., 2018a; Jiang et al., 2018b). Jiang et al. (2014, particle size (< 2 mm fractions) distribution was measured by a laser
2018b) reported that the runoff volume and sediment yield increased diffraction device (Better Size Ltd., Dandong, Liaoning Province,
with rainfall intensity and that unit stream power was the optimal China). The main chemical and physical properties of the soil samples
hydrodynamic parameter by which to characterise the dynamic me- are summarised in Table 1. The gravel was passed through 3, 5 and
chanisms of rill erosion. In a recent study, the optimised flow velocity 10 mm sieves. The gravel sizes of 2–3, 3–5, and 5–10 mm represented
was found to predict Tc (Jiang et al., 2018a). Moreover, colluvial mass proportions of 40, 45 and 15%, respectively.

2
Z. Zhan, et al. Catena 188 (2020) 104447

Upper catchment Collapsing wall

A B

Alluvial fan Colluvial deposit


Fig. 1. A typical collapsing gully in the study area, locally referred to as a benggang (A): a collapsing gully with an upper catchment, collapsing wall, colluvial deposit,
and alluvial fan. (B) Example of loose colluvial deposits from the upper catchment and collapsing wall.

2.3. Experimental setup to the designated values before simulation of the Tc.

The experimental setup mainly consisted of a hydraulic flume,


2.4. Simulations of sediment transport capacity
water supply system and sediment feeding hopper (Fig. 3). The flume
channel (Fig. 3A) was 4 m long (L), 0.12 m wide (W) and 0.10 m high
2.4.1. Experimental design
(H) and was made of a zincified sheet iron plate. The bed slope of the
To study the effect of gravel content on the Tc of overland flow, the
flume could be adjusted from 0 to 84%. The L × W × H of the ex-
fine soil and gravel samples were thoroughly mixed together to six
perimental rill flume were similar to those described by Zhang et al.
specified gravel mass fractions (C) based on the colluvial deposits with
(2009) and Wu et al. (2016). Flow discharge was controlled by a
gravel contents exceeding 40%: 0 (no gravel), 10, 20, 30, 40, and 50%,
peristaltic pump (Fig. 3F) (model WT600-4F-C; Longer Pump Ltd.,
with the sediment median grain sizes (d50) being 0.05, 0.06, 0.11, 0.39,
Baoding, Hebei Province, China) and measured directly by a calibrated
0.78, and 2.00 mm, respectively. The different gravel contents were
flow control meter. The sediment feeding rates were controlled accu-
based on the percentage of gravel weight relative to the total mass.
rately and stably by a conveyor (Fig. 3D) atop the hopper (Fig. 3C). For
Thus, a 50% gravel content represents 50% gravel weight and 50% soil
each treatment, the slope gradients and flow discharges were adjusted
weight or a gravel:soil weight ratio of 1:1.

Fig. 2. Location of the research site.

3
Z. Zhan, et al. Catena 188 (2020) 104447

Table 1
Major chemical and physical properties of the tested soils.
Organic matter (g kg−1) *CEC (cmol kg−1) pH Sand 0.05–2 mm (%) Silt 0.002–0.05 mm (%) Clay < 0.002 mm (%)

1.68 ± 0.32 1.07 ± 0.02 5.19 ± 0.07 50.46 44.51 5.03

Note: *CEC: Cation-exchange capacity, ± : standard deviation.

The colluvial deposits had a maximum slope of 84% based on field 2014; Lin et al., 2017).
observations. The slope gradients (S) were fixed at 18, 36, 58, and 84%
(the tangent values of the slope were 10°, 20°, 30°, and 40°, respec- 2.4.2. Experimental procedure
tively). The flow discharge per unit width (q) for each set of experi- The experiments were conducted at the Science and Technology
ments was fixed at 0. 56, 1.11, 2.22, and 4.44 × 10-3 m2 s−1 (4, 8, 16, Research Center for Soil and Water Conservation at Fujian Agriculture
and 32 L min−1, respectively). The experiment was designed on the and Forestry University (FAFU), Fuzhou city, Fujian Province, south-
basis of local field survey and precipitation record data (Jiang et al., eastern China. A 5-mm layer of the soil with 30% gravel content was

Baffle

Adjustable end stop C F


Hopper Peristaltic pump
D
Inlet Water pipe
Conveyor

30 ccm
10 cm

E G
20 cm

Open box 10 cm Water pool

15 cm

H
Zincified sheer iron plate bed with sediment

Outlet
Over-flow tank

Sample bucket

Hopper
Open box Flume Peristaltic pump

Conveyor
Fig. 3. Experimental equipment. The above picture is a schematic diagram of the experimental setup, and the following picture is the experimental device.

4
Z. Zhan, et al. Catena 188 (2020) 104447

glued on the flume bed to simulate grain roughness and remained ∑ (Oi − Pi )2
NSE = 1 −
constant during the experiments. Two sediment sources with gravel ∑ (Oi − O¯ )2 (4)
were designed to ensure that the proper Tc was attained for each of the 2
different experimental conditions. The first sediment source used the where r is the coefficient of determination; RE is the relative error; Oi is
conveyor (Fig. 3D) to fill the hopper over the flume at a distance of the measured value; Pi is the predicted value; Ō is the mean of the
0.3 m from the upper inlet. The sediment feeding rate was controlled by observed values; P̄ is the mean of the predicted values; and NSE is the
the speed of the conveyor and could be adjusted using an electronic Nash-Sutcliffe efficiency (Nash and Sutcliffe, 1970). NSE can evaluate
control, which was set at the beginning of each test and remained fixed the degree of fit of an equation between the predicted and measured
during the test. A second sediment source involved a 20 × 12 × 10 cm values. Generally, the NSE value ranges from minus infinity to 1. When
(L × W × H) open box placed in the flume bed with the 0.5 m lower NSE ≤ 0.40, the equation simulation results are unsatisfactory; when
edge of the box located above the outlet of the flume (Fig. 3E); this 0.4 < NSE ≤ 0.7, the equation simulation results are satisfactory; and
sediment source was used as a supplement for testing the Tc. According when NSE > 0.7, the equation simulation results are good (Ahmad
to erosion theory, if Tc was not reached due to insufficient feeding from et al., 2011).
the hopper, the deficit would be compensated for by sediment en-
trainment from the box to reach the Tc (Zhang et al., 2009; Wang et al., 2.6. Statistical analysis
2015; Wu et al., 2016). Before the experiment, test sediments with 15%
soil moisture were evenly filled in the box and then covered with steel Graphs and tables were constructed using Origin 8.0 and Excel
plates. After the flow in the flume was stable, the first sediment source 2010, respectively. An analysis of variance (ANOVA) was used to
was introduced. During the test, a wire rod was used to continuously analyse the effects of the slope gradient, flow discharge and gravel
stir the sediment and set the sediment in motion under the hopper. The content on the Tc. The relationships were considered significant at the
sediment feeding rate increased gradually until the fed sediment could 0.05 significance level, as examined using SPSS (Statistical Product and
not be completely transported, as indicated by the deposition of sedi- Service Solutions) software (version 18.0). A regression analysis was
ment immediately downstream from the first sediment feeder. The steel employed to estimate the relationships between the slope gradient, flow
sheet above the open box was removed after which the measurements discharge, and gravel content or sediment size.
started. One water-sediment mixture sample was collected at the outlet
at regular intervals, and the sampling period was recorded using a di- 3. Results and discussion
gital stopwatch and adjusted depending on flow discharge (longer for
small q values and shorter for larger q values) within the range of 3.1. Effects of slope gradient and flow discharge on sediment transport
5–10 s. Six samples of water-sediment mixture were collected using capacity
plastic buckets for each test. The water-sediment mixtures were mea-
sured via the oven-drying method. The dry sediment weight was di- The data in Fig. 4 show that the slope gradient affected the mea-
vided by the sampling time and the flume width to obtain the Tc (Zhang sured Tc under a constant gravel content and flow discharge. The
et al., 2009). The average of six samples was used as the measured measured Tc varied from 0.26 ± 0.03 to 7.18 ± 0.40 kg m−1 s−1. The
equilibrium Tc for a given combination of flow discharge and slope Tc values increased with increasing slope gradient. Specifically, the
gradient. change in sediment transport Tc exhibited a slow increase followed by
stabilisation when the slope gradient increased from 58% to 84%.
Furthermore, the relationship between Tc and slope gradient was de-
2.5. Data analysis scribed by a power equation, and the exponents of the slope gradient
ranged from 0.194 to 0.626, with a mean value of 0.383 and r2 > 0.85.
To determine whether the changes in Tc for two adjacent gravel Wang et al. (2015) reported exponential functions that could describe
contents were regular, the percent increase in Tc between two adjacent the relationships between the Tc and slope gradient, the equations were
gravel contents for different slope gradients and flow discharges was different from those obtained in this study. This difference indicated
calculated by the following equation: that different experimental conditions resulted in different relationships
Tc (i) − Tc (i − 10) between Tc and slope gradient.
P= × 100% Fig. 5 reveals that the measured Tc varied with flow discharge for a
Tc (i − 10) (1)
constant slope gradient and gravel content. The Tc increased as the flow
where P represents the percent increase in Tc (%); Tc(i) is the Tc of the discharge increased. For the same level of flow discharge, the increase
gravel content; and i is the group number. Gravel contents of 10, 20, 30, in Tc was largest when the slope gradient ranged from 36% to 84%
40 and 50% at slope gradients of 18, 36, 58, and 84% and flow dis- (Fig. 5). This result could be attributed to the presence of a critical slope
charges of 0.56, 1.11, 2.22, and 4.44 × 10-3 m2 s−1 were included. (> 36%) for transporting colluvial soil containing gravel. Similarly, Wu
The data set of sediment contents was analysed to derive equations et al. (2016) and Jiang et al. (2018a) also found a critical slope for
that could describe the relationships of the Tc with the slope gradient, transporting a different soil and attributed the relationship to gradual
flow discharge and gravel content. The data set was also used to vali- increases in flow kinetic energy and sediment potential energy with an
date the equations by generating values of the statistical parameters r2, increase in the slope gradient. However, the magnitude of the increase
relative error (RE) and Nash-Sutcliffe efficiency (NSE), which were used in these parameters decreased gradually because the increase in the
to evaluate the performance of the established equations. The r2, RE and sine component (the difference between the sine of the steep slope and
NSE values were calculated as follows: that of the shallow slope) decreased gradually as the slope gradient
increased.
n 2
⎡∑ (O − O¯ )(P − P¯ ) ⎤ Multivariate regression analyses were conducted to evaluate the
⎢i i i
⎥ relationships of the Tc with the slope gradient and flow discharge under
r = n⎣
2
n

six gravel contents in this study. The Tc increased following a power
¯
∑ (Oi − O) ∑ (Pi − P¯ )2
2
function with the slope gradient and flow discharge for a range of
i=1 i=1 (2)
gravel contents. The exponents of the slope gradients ranged from
0.266 to 0.295, with a mean value of 0.278, and the exponents of the
Pi − Oi
RE = × 100% flow discharges ranged from 0.943 to 1.041, with a mean value of
Oi (3)
0.993. However, the changes in the power function trend were not

5
Z. Zhan, et al. Catena 188 (2020) 104447

Fig. 4. Measured Tc as a function of slope gradient under a constant flow discharge and gravel content.

consistent with the increasing gravel contents (Table 2). The exponents discharge was a power equation:
of the flow discharge were much greater than those of the slope gra-
Tc = aSbqc (5)
dient. This finding indicated that the Tc was more strongly influenced
by the flow discharge in this study. Hai et al. (2017) drew similar where Tc is the sediment transport capacity (kg m−1 s−1); S is the slope
conclusions in which the exponents of the slope gradient and flow gradient (m m−1); q is the flow discharge (m2 s−1); and a, b, and c are
discharge were 0.511 and 0.854, respectively, and Tc was more sensi- regression coefficients.
tive to unit discharge than to the slope gradient on a steep loessial soil Prosser and Rustomji (2000) recommended the values of
slope with an erodible bed. 0.9 ≤ b ≤ 1.8 and 1.0 ≤ c ≤ 1.8 for Tc modelling when Eq. (5) is used.
The relationship of the Tc with the slope gradient and flow discharge The results of this study showed that the exponents of flow discharge
could be described by a power function for all gravel contents, with r2 ranging from 0.943 to 1.041 were nearly within their minimum range
values ranging from 0.980 to 0.996, with a mean value of 0.989, and (1.0 ≤ b ≤ 1.8). However, the exponents of the slope gradients ranged
NSE values ranging from 0.980 to 0.996, with a mean value of 0.988. from 0.266 to 0.295, which were smaller than the slope gradient values
These results indicated that slope gradient and flow discharge were (0.9 ≤ c ≤ 1.8), consistent with the conclusions of Jiang et al. (2018a)
good predictors of Tc and should be selected as simulation parameters. for colluvial soil in a nonerodible bed. This finding, in which the ex-
Fig. 6 shows that the RE changed with the flow discharge across a range ponents of the slope gradient were less than those of the flow discharge,
of slope gradients for six gravel contents. The RE values varied from indicated that similar experimental conditions yielded similar results.
−10.58 to 67.51%. However, the mean RE percentage in Fig. 6 was The reason for this effect was that the increase in the sine component
approximately 0 as the gravel content increased; the mean RE percen- declined gradually with an increase in the slope gradient, such that the
tages becoming approximately 0 indicated that the predicted Tc was kinetic energy of the rill flow and the potential energy of the sediment
extremely close to the measured values. From the r2 (> 0.98) and NSE first increased and then gradually plateaued (Jiang et al., 2018a). Si-
values (> 0.98) shown in Table 2, we can infer that the expressions milarly, the sinusoidal component of the kinetic energy of the rill flow
predicted Tc extremely well. and the potential energy of the sediment decreased with the increase in
Prosser and Rustomji (2000) summarised previous studies and slope gradient. The slopes (18–84%) implemented in this study were
suggested that the relationship among the Tc, slope gradient and flow much steeper than the gentle slopes (< 10%) used in the equation by

6
Z. Zhan, et al. Catena 188 (2020) 104447

Fig. 5. Measured Tc as a function of flow discharge under a constant slope gradient and gravel content.

Table 2
Tc as a power function of the slope gradient and flow discharge for six gravel
contents.
Gravel content (%) Equation r2 NSE p n

0.292 1.024
0 Tc = 1206.07S q 0.996 0.996 0.01 16
10 Tc = 841.66S0.266q0.943 0.980 0.980 0.01 16
0.295 0.954
20 Tc = 1011.92S q 0.986 0.986 0.01 16
30 Tc = 1551.23S0.241q1.016 0.987 0.987 0.01 16
40 Tc = 2059.26S0.286q1.041 0.992 0.991 0.01 16
50 Tc = 1555.57S0.288q0.981 0.991 0.991 0.01 16

Where Tc is the sediment transport capacity (kg m−1 s−1); q is the flow dis-
charge (m2 s−1); S is the slope gradient (m m−1).

Prosser and Rustomji (2000). Therefore, Prosser and Rustomji did not
recommend the empirical exponents of slope gradients for modelling
the Tc of colluvial deposits. Fig. 6. A box-plot of REs for different gravel contents. Each box indicates the
25th/75th percentiles. The whisker caps represent the 10th/90th percentiles.
3.2. Effects of gravel content on sediment transport capacity The median is depicted by the line and the mean value is depicted by the
square.
Fig. 7 shows that gravel content affected the measured Tc under a
constant slope gradient and flow discharge. Overall, the gravel content enhanced Tc (Wang and Shi, 2015; Jiang et al., 2018a). On the other
was directly proportional to Tc. When the slope gradient and flow dis- hand, the bulk density and buoyancy of the hyperconcentrated flow
charge were unchanged, high gravel contents promoted sediment were high (the sediment concentrations were all greater than
transport. Sediment with a high gravel content is easier to transport 370 kg m−3 in this test), causing a reduction in the settling rate of the
under the action of gravity and inertia on steep slopes, resulting in an sediment particles. Furthermore, the consumption of the particle

7
Z. Zhan, et al. Catena 188 (2020) 104447

Fig. 7. Measured Tc as a function of gravel content under a constant slope gradient and flow discharge.

suspension and turbulent energy by the flow was reduced; conse- fine soil. With an increase in the slope gradient and flow discharge, the
quently, the Tc of the flow was enhanced and more coarse particles obstructive effect of the gravel gradually decreased or disappeared, and
could be transported (Xu, 1999). However, Zhang et al. (2011) de- the gravity and inertia effects of the gravel itself promoted the increase
monstrated that the measured Tc decreased as the sediment particle size in flow velocity. The capacity of the overland flow was sufficient to
increased. In their study, the sediment size of the fine soil ranged from transport the mixture of gravel and fine soil in this case. Notably, this
0.02 to 2.00 mm without gravel, and the slope gradient ranged from 9 phenomenon became more obvious as the slope gradient and flow
to 47%. The authors concluded that the larger the sediment size is, the discharge increased.
greater the energy needed for setting the sediment in motion. There- As shown in Fig. 7, the measured Tc for different gravel contents at a
fore, the above discussion indicates that the effects of sediment size on constant slope gradient and flow discharge was fitted by a linear
Tc are complex and should be studied further. equation:
As shown in Fig. 7, the measured Tc increased as the gravel content
Tc = kC + u (6)
increased. However, different gravel contents exerted different effects
−1 −1
on Tc. The increase in Tc remained almost unchanged as the gravel where Tc is the sediment transport capacity (kg m s ); k is the slope
content increased, and the variations tended to become nearly parallel of the fitted line; u is the intercept; and C is the gravel content (%).
to the axis of the gravel content when the flow discharge reached Table 3 summarises the fitted parameters obtained using Eq. (6)
0.56 × 10-3 m2 s−1, which suggested that Tc might not increase sig- under different slope gradients and flow discharges. Table 3 demon-
nificantly above this flow discharge because the presence of gravel strates that the Tc increased linearly in response to the increasing gravel
caused the flow to be slow and stable. Similarly, the sediment transport content. Nearly all k and u parameters increased as the slope gradient
rate and sediment yield decreased in response to the increasing gravel increased, and consistent trends among different slope gradients and
content in the studies of Wilcock et al. (2001) and Luo et al. (2019). flow discharges were observed. The parameter k ranged from 0.003 to
However, the increase in Tc with the increase in the flow discharge was 0.060 (Table 3). Higher k values indicate a steeper slope, representing a
more striking, as flow has more energy to transport sediment. In ad- more rapid increase in Tc with an increasing gravel content. The k value
dition, Fig. 7B and C show that the variation in Tc with increasing increased as the flow discharge increased when the slope gradient re-
gravel content on gentle slopes (18–36%) was less than that on steep mained unchanged. The measured Tc also increased rapidly as the flow
slopes (58–84%) at a low flow discharge (0.56 × 10-3 m2 s−1). discharge increased. Lower k values indicated that the effect of the
Nevertheless, as shown in Fig. 7B and D, this variation in Tc was clearly gravel content on Tc slightly increased as the flow discharge increased
lower at a slope gradient of 18% than on the other slopes when the flow for each slope gradient. The parameter u ranged from 0.25 to 4.19
discharge exceeded 1.11 × 10-3 m2 s−1. This finding indicated the (Table 3), and the u values increased as the slope gradient increased at
formation of slow and stable flow because gravel had an obstructive each flow discharge. The coefficient k and its variation in each case in
effect on overland flow at flow discharges of 0.56 and 1.11 × 10- Table 3 could also explain the variation in Tc in Fig. 7.
3
m2 s−1, and hence, the energy of this flow could only transport mostly Table 4 lists the percent increases in Tc between two adjacent gravel
contents by Eq. (1) for different slope gradients and flow discharges.

8
Z. Zhan, et al. Catena 188 (2020) 104447

Table 3 content (p < 0.01). Nonlinear regression analysis indicated that Tc


The parameters k, u and r2 observed for different slope gradients and flow could be estimated by a power equation either with slope gradient, flow
discharges. discharge and gravel content, or with slope gradient, flow discharge
Flow discharge (10−3 m2 s−1) Slope gradient (%) Parameter and sediment median grain size (d50) for the pooled data set:

k u r2 Tc = 1090S 0.268q0.980C 0.052 (r 2 = 0.96, NSE = 0.96, n = 96) (7)

0.56 18 0.003 0.25 0.98 Tc = 1439S 0.268q0.980d50 0.117 (r 2 = 0.98, NSE = 0.98, n = 96) (8)
36 0.004 0.28 0.95
−1 −1
58 0.006 0.37 0.97 where Tc is the sediment transport capacity (kg m s ); S is the slope
84 0.005 0.47 0.98 gradient (m m−1); q is the flow discharge (m2 s−1); C is the gravel
1.11 18 0.004 0.56 0.95 content (%), where if C is 0, C + 0.01 is recommended to replace the C
36 0.014 0.90 0.96 value in Eq. (7); and d50 is the sediment median grain size (mm).
58 0.015 1.00 0.98 Comparisons between the measured and predicted Tc values using
84 0.018 1.04 0.92
Eq. (7) and Eq. (8) are plotted in Fig. 8. Eq. (7) simulated the measured
2.22 18 0.015 1.33 0.93 Tc adequately with r2 = 0.96, RE ranging from −21.27 to 85.01%, with
36 0.022 2.01 0.95
a mean RE of 13.00%, and thus could be used to accurately predict Tc,
58 0.026 2.25 0.90
84 0.030 2.39 0.97 as indicated by NSE = 0.96 > 0.7. Eq. (8) also simulated the measured
Tc adequately with r2 = 0.98, RE ranging from −18.98 to 73.91%, with
4.44 18 0.430 2.79 0.96
36 0.470 3.57 0.97 a mean RE of 11.53%, and thus could also be used to predict Tc accu-
58 0.520 3.85 0.98 rately, as indicated by NSE = 0.98 > 0.7. As shown in Fig. 8, the
84 0.600 4.19 0.97 predicted Tc showed good agreement with the measured values, and the
plotted data were all close to the 1:1 line. The predicted Tc showed a
Note: k is the slope of the fitted line, and higher k values indicate a steeper
good linear relationship with the measured values. Therefore, Eq. (7)
slope, representing a more rapid increase in the Tc with the gravel content; u is
and Eq. (8) could facilitate prediction of the Tc of colluvial deposits in
the intercept.
this study. However, at a flow discharge of 0.56 × 10-3 m2 s−1, the Tc
values calculated by Eqs. (7) and (8) were greater than the measured
Table 4
values for a measured Tc < 1 kg m−1 s−1 (Fig. 8). The Tc values
Percent increases in the Tc of adjacent gravel contents for constant slope gra-
dients and flow discharges.
calculated by Eq. (7) and Eq. (8) were smaller than the measured values
when the measured Tc ranged from 2 to 4 kg m−1 s−1 with a flow
Flow discharge Gravel Increased proportion of Tc at slope gradient discharge of 2.22 × 10-3 m2 s−1 (Fig. 8). The Tc was overestimated
(10-3 m2 s−1) content (%) (%)
because the roughness of the experimental flume greatly influenced
18% 36% 58% 84% flow resistance and impeded gravel movement; as a result, the flow
could only carry mostly fine particles at a flow discharge of 0.56 × 10-
0.56 0–10 11.1 16.5 3.3 15.5 3
m2 s−1. Conversely, the small effect of the roughness on the flow
10–20 4.4 8.0 31.7 7.0
20–30 5.3 10.1 3.9 9.0
resistance resulted in an overestimation of the Tc when the flow dis-
30–40 9.0 9.4 13.9 13.0 charge was 2.22 × 10-3 m2 s−1. The results of comparisons between the
40–50 12.5 31.1 2.8 4.1 measured and predicted Tc using equations established in different test
1.11 0–10 5.8 23.4 15.5 11.3 conditions revealed differences. Zhang et al. (2011) reported that the
10–20 6.2 9.9 17.4 13.7 model (Tc = 2382.32S1.637q1.269d50-0.345) overpredicted Tc when the Tc
20–30 2.6 3.6 5.3 8.3 values were greater than 7 kg m−1 s−1. The Tc values calculated by
30–40 6.1 16.4 10.4 9.4 equation (Tc = 0.17 × 106S2.89q1.46d50-0.50) in the case of Ali et al.
40–50 10.2 13.4 14.1 25.2
(2012) were greater than the measured values when the measured
2.22 0–10 14.2 25.3 32.7 23.7 Tc < 0.003 kg m−1 s−1. This inconsistent prediction further revealed
10–20 7.9 4.9 4.9 9.7
20–30 4.9 7.6 4.1 8.4
limitations for this type of regression model in predicting Tc.
30–40 6.9 6.6 10.0 9.4 The exponents of the slope gradient, flow discharge and gravel
40–50 18.3 9.1 6.7 5.5 content or d50 were 0.268, 0.990 and 0.052 or 0.117, respectively; thus,
4.44 0–10 16.4 9.6 8.5 6.2 Tc was more sensitive to flow discharge than to the slope gradient, and
10–20 6.5 11.3 5.8 13.6 Tc was most weakly influenced by the gravel content or d50. Many
20–30 22.8 16.2 19.4 8.9 studies have shown that the slope gradient and flow discharge were the
30–40 8.9 5.5 9.0 8.1 main factors restricting the variation in Tc, and these parameters have
40–50 3.7 6.4 4.1 15.1
often been used as the most basic variables for studying Tc; furthermore,
those studies came to the same conclusion that the slope gradient af-
The Tc percentage increased from 2.6 to 32.7%, and the increments of fected Tc more than flow discharge (Ali et al., 2012; Beasley and
the Tc for the two adjacent gravel contents were both positive. How- Huggins, 1982; Mu et al., 2019; Polyakov and Nearing, 2003; Wang and
ever, the changes in the Tc in the two adjacent gravel contents were not Shi, 2015; Zhang et al., 2011). However, that conclusion is contra-
obviously regular. The above analysis showed that the variation in Tc dictory to our findings. Although the effect of gravel content on Tc was
between two adjacent gravel contents was indeterminate. The cause of less than those of the slope gradient and flow discharge, gravel content
this effect is unknown, and further studies are needed to investigate it. played an important role in the process of Tc. Moreover, Zhang et al.
(2011) and Ali et al. (2012) also reported that the effect of d50 on Tc was
weaker than the effects of the slope gradient and flow discharge on
3.3. Combined effects of slope gradient, flow discharge and gravel content sediments from the bed of the Yongding River near Beijing in a none-
on sediment transport capacity rodible bed and on noncohesive soils in an erodible bed, respectively.
The principal difference was that the correlation between d50 and Tc
In the previous analysis, the measured Tc values were found to in- was negative in their study (the exponents of d50 were −0.345 and
crease with the slope gradient, flow discharge and gravel content. Tc −0.50, respectively), but positive in this research (the exponent of d50
was greatly affected by the slope gradient, flow discharge and gravel was 0.117). The reason for this difference was that the sediment

9
Z. Zhan, et al. Catena 188 (2020) 104447

Fig. 8. Measured vs. predicted Tc values across a range of flow discharges using Eq. (7) (A) and Eq. (8) (B).

particles in the test materials in the previous study were less than 2 mm flow discharge and gravel content on the Tc of overland flow on col-
in diameter, whereas the gravel sizes used in Eq. (7) ranged from 2 to luvial deposits in southeastern China. The results suggested that the Tc
10 mm. Furthermore, the experimental slope setting was also different. increased according to a power function with an increasing slope gra-
A maximum slope of 84% was used in this study, resulting in a range dient, and Tc exhibited a slow increase followed by stabilisation. Tc
that was 2 to 5 times wider than that in preceding studies (e.g., the increased linearly in response to increasing flow discharge. Tc increased
maximum slope values in Zhang et al. (2011) and Ali et al. (2012) were with the increase in gravel content. However, the increase in Tc as the
42.3% and 17.6%, respectively). Additionally, the effect of different gravel content increased at a low flow discharge of 0.56 × 10-3 m2 s−1
erosional beds on the experimental results should also be taken into was slower than that at a flow discharge ranging from 1.11 to
account. The equation established by Zhang et al. (2011) indicated that 4.44 × 10-3 m2 s−1. The variation in Tc with increasing gravel content
Tc decreased with increasing sediment size because the larger the se- on gentle slopes was smaller than that on steep slopes. The slope gra-
diment size, the more energy is needed to set the sediment in motion. dient, flow discharge and gravel content all had extremely significant
Conversely, compared with sediment less than 2 mm, gravel has a effects on Tc, which increased as a power function with the slope gra-
greater gravitational potential energy and inertia on steep slopes and dient and flow discharge at six gravel contents. Tc can be estimated with
can roll more easily (Wang and Shi, 2015), resulting in the gravel a power equation involving either slope gradient, flow discharge and
content having the weakest effect on the Tc and the exponent of the gravel content or slope gradient, flow discharge and sediment median
gravel content or sediment size being smaller than the exponents grain size (d50). The two functions overestimated Tc when the measured
measured in other studies. Consequently, Eq. (7) or Eq. (8) based on the Tc was less than 1 kg m−1 s−1, but underestimated Tc when the mea-
gravel content or sediment median grain size on colluvial deposits sured Tc ranged from 2 to 4 kg m−1 s−1. These integrated results in-
provided a suitable method for the accurate estimation of Tc. dicate that the two new power functions established in this study can
facilitate the prediction of Tc in colluvial deposits.
3.4. Implications of findings for regional landscape scale
Declaration of Competing Interest
In this study, a laboratory simulation of Tc was employed to quantify
the effects of gravel content on Tc. The empirical formulas with the The authors declare that they have no known competing financial
gravel content to predict Tc were established. Therefore, for the varia- interests or personal relationships that could have appeared to influ-
tion in Tc on colluvial deposits in regional landscape scales, one way is ence the work reported in this paper.
to collect the local rainfall data, measure the range of slope gradient
and acquire the spatial distributions of gravel content on colluvial de- Acknowledgements
posits, and use the empirical formulas to obtain the variation in Tc of
overland flow. In addition, the empirical formulas can also be used to We gratefully acknowledge the funding provided by the National
predict Tc in soil and rock mixtures composed of different types or Natural Science Foundation of China (No: 41977071), the National
particles (such as engineering accumulations, purple soil, etc.). Natural Science Foundation of China (No: 41571272), the Fujian
Provincial Natural Science Foundation Projects (No: 2017J01459), and
3.5. Limitations of approach and improvements for future research the Co-Innovation Center For Soil and Water Conservation in Red Soil
Region of the Cross-Straits (No: K80ND800303). We also thank
The variation in the Tc between two adjacent gravel contents was Haidong Zhang, Xiaojun Yan, Jiaxiang Cai, Xinbo Li and Jian Zhou for
indeterminate (Table 4). This might be due to the reasons that the ex- their assistance with sample analysis.
perimental gravel contents ranged from 0 to 50% on colluvial deposits,
which did not reach the earlier reported maximum gravel content of References
70% (Jiang et al., 2014). In future, a larger range of gravel contents
should be used to study the variation in Tc. On the other hand, the Abrahams, A.D., Li, G., Krishnan, C., Atkinson, J.F., 2001. A sediment transport equation
application of the empirical formulas on different types of soil merited for interrill overland flow on rough surfaces. Earth Surf. Proc. Land. 26 (13),
1443–1459.
and the relationship of hydrodynamic variables with Tc on colluvial Ahmad, H.M.N., Sinclair, A., Jamieson, R., Madani, A., Hebb, B., Havard, P., Yiridoe, E.K.,
deposits need further verification. 2011. Modeling sediment and nitrogen export from a rural watershed in eastern
Canada using the soil and water assessment tool. J. Environ. Qual. 40 (4),
1182–1194.
4. Conclusions Aksoy, H., Eris, E., Tayfur, G., 2016. Empirical sediment transport models based on indoor
rainfall simulator and erosion flume experimental data. Land Degrad. Dev. 28 (4),
This study was conducted to investigate the effect of slope gradient, 1320–1328.

10
Z. Zhan, et al. Catena 188 (2020) 104447

Agarwal, A., Dickinson, W.T., 1991. Effect of texture, rainfall and slope on rainfall in- colluvial deposits in a permanent gully. Catena 157, 47–57.
terrill sediment transport. Nord. Hydrol. 22, 227–242. Low, H.S., 1989. Effect of sediment density on bed-load transport. J. Hydraul. Eng. 115
Ali, M., Sterk, G., Seeger, M., Boersema, M.P., 2012. Effect of hydraulic parameters on (1), 124–138.
sediment transport capacity in overland flow over erodible beds. Hydrol. Earth Syst. Luo, H., Rong, Y., Lv, J., Xie, Y., et al., 2019. Runoff erosion processes on artificially
Sci. 16 (4), 591–601. constructed conically-shaped overburdened stockpiles with different gravel contents:
Beasley, D.B., Huggins, L.F., 1982. ANSWERS user’s manual. Department of Agriculture Laboratory experiments with simulated rainfall. Catena 175, 93–100.
Engineers, Purdue University, West Lafayette, IN. Mahmoodabadi, M., Ghadiri, H., Rose, C., Yu, B.F., Rafahi, H., Rouhipour, H., 2014.
Brakensiek, D.L., Rawls, W.J., 1994. Soil containing rock fragments effects on infiltration. Evaluation of GUEST and WEPP with a new approach for the determination of se-
Catena 23 (1–2), 99–110. diment transport capacity. J. Hydrol. 513, 413–421.
Cerdà, 2001. Effects of rock fragment cover on soil infiltration, interrill runoff and ero- Miller, F.T., Guthrie, R.L., 1984. Classification and distribution of soils containing rock
sion. Eur. J. Soil Sci. 2010, 52, 59–68. fragments in the United States. Soil Sci. Soc. Am. J., Spec. Publ.
Chen, J.L., Zhou, M., Lin, J.S., Huang, B.F., Xu, T.T., Wang, M.K., Ge, H.L., Huang, Y.H., Mu, H.L., Yu, X.J., Fu, S.H., Yu, B.F., Liu, Y.N., Zhang, G.H., 2019. Effect of stem basal
2018. Comparison of soil physicochemical properties and mineralogical compositions cover on the sediment transport capacity of overland flows. Geoderma 337, 384–393.
between noncollapsible soils and collapsed gullies. Geoderma 317, 56–66. Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part
De Roo, A.P.J., 1996. LISEM: a single-event physically based hydrological and soil erosion I—A discussion of principles. J. Hydrol. 10 (3), 282–290.
model for drainage basins. I: theory, input and output. Hydrol. Process. 10, Nearing, M.A., Foster, G.R., Lane, L.J., Finkner, S.C., 1989. A process-based soil erosion
1107–1117. model for USDA-Water erosion prediction project technology. Trans. Am. Soc. Agric.
Dong, X., Ding, S.W., Li, L., Deng, Y.S., Wang, Q.X., Wang, S.L., Cai, C.F., 2016. Effects of Eng. 35, 1587–1593.
collapsing gully erosion on soil qualities of farm fields in the hilly granitic region of Nearing, M.A., Norton, L.D., Bulgakov, D.A., Larionov, G.A., West, L.T., Dontsova, K.M.,
south china. J. Integr. Agric. 15 (12), 2873–2885. 1997. Hydraulics and erosion in eroding rills. Water Resour. Res. 33 (4), 865–876.
Figueiredo, T.D., Poesen, J., 1998. Effects of surface rock fragment characteristics on Poesen, J., Torri, D., Bunte, K., 1994. Effects of rock fragments on soil erosion by water at
interrill runoff and erosion of a silty loam soil. Soil Tillage Res. 46 (1–2), 81–95. different spatial scales: a review. Catena 23, 141–166.
Finkner, S.C., Nearing, M.A., Foster, G.R., Gilley, J.E., 1989. Simplified equation for Polyakov, V.O., Nearing, M.A., 2003. Sediment transport in rill flow under deposition and
modeling sediment transport capacity. Trans. Am. Soc. Agric. Eng. 32 (5), detachment conditions. Catena 51 (1), 33–43.
1545–1550. Prosser, I.P., Rustomji, P., 2000. Sediment transport capacity relations for overland flow.
Gong, J.G., Jia, Y.W., Zhou, Z.H., Wang, Y., Wang, W.L., Peng, H., 2011. An experimental Prog. Phys. Geogr. 24 (2), 179–193.
study on dynamic processes of ephemeral gully erosion in loess landscapes. Soil Survey Staff, 2014. Keys to Soil Taxonomy, 12th Ed. United States Department of
Geomorphology 125, 203–213. Agriculture and Natural Resources Conservation Service, Washington, D. C.
Govers, G., 1990. Empirical relationships on the transporting capacity of overland flow. Wang, L., Shi, Z.H., 2015. Size selectivity of eroded sediment associated with soil texture
Publ. Int. Assoc. Hydrol. Sci. Publ. 189, 45–63. on steep slopes. Soil Sci. Soc. Am. J. 79 (3), 917–929.
Jackson, M.L., 1979. Soil Chemical Analysis, second ed. University of Wisconsin, Wang, Z.L., Yang, X., Liu, J., Yuan, Y., 2015. Sediment transport capacity and its response
Madison, WI. to hydraulic parameters in experimental rill flow on steep slope. J. Soil Water
Jiang, F.S., Huang, Y.H., Wang, M.K., Zhao, G., Ge, H.L., 2014. Effects of rainfall intensity Conserv. 70 (1), 36–44.
and slope gradient on steep colluvial deposit erosion in Southeast China. Soil Sci. Soc. Wilcock, P.R., Kenworthy, S.T., Crowe, J.C., 2001. Experimental study of the transport of
Am. J. 78 (5), 1741–1752. mixed sand and gravel. Water Resour. Res. 37 (12), 3349–3358.
Jiang, F.S., Gao, P.Y., Si, X.J., Zhan, Z.Z., Zhang, H.D., Lin, J.S., Ji, X., Wang, M.K., Wu, B., Wang, Z.L., Shen, N., Wang, S., 2016. Modelling sediment transport capacity of
Huang, Y.H., 2018a. Modelling the sediment transport capacity of flows in steep rill flow for loess sediments on steep slopes. Catena 147, 453–462.
nonerodible rills. Hydrol. Process. 32, 3852–3865. Xu, J., 1996. Benggang erosion: The influencing factors. Catena 27, 249–263.
Jiang, F.S., Zhan, Z.Z., Chen, J.L., Lin, J.S., Wang, M.K., Ge, H.L., Huang, Y.H., 2018b. Rill Xu, J.X., 1999. Erosion caused by hyperconcentrated flow on the loess plateau of China.
erosion processes on a steep colluvial deposit slope under heavy rainfall. Catena 169, Catena 36, 1–19.
46–58. Zhong, B., Peng, S., Zhang, Q., Ma, H., Cao, S., 2013. Using an ecological economics
Hai, X., Liu, G., Liu, P., Zhen, F.L., Zhang, J.Q., Hu, F.N., 2017. Sediment transport ca- approach to support the restoration of collapsing gullies in southern China. Land Use
pacity of concentrated flows on steep loessial slope with erodible beds. Sci. Rep. 7 (1), Policy 32, 119–124.
2350. Zhang, G.H., Liu, Y.M., Han, Y.F., Zhang, X.C., 2009. Sediment transport and soil de-
Lei, T.W., Zhang, Q., Zhao, J., Tang, Z., 2001. Laboratory study of sediment transport tachment on steep slopes: I. Transport capacity estimation. Soil Sci. Soc. Am. J. 73
capacity in the dynamic process of rill erosion. Trans. Am. Soc. Agric. Eng. 44 (6), (4), 1291–1297.
1537–1542. Zhang, G.H., Wang, L.L., Tang, K.M., Luo, R.T., Zhang, X.C., 2011. Effects of sediment size
Lin, J.S., Huang, Y.H., Zhao, G., Jiang, F.S., Wang, M.K., Ge, H.L., 2017. Flow-driven soil on transport capacity of overland flow on steep slopes. Int. Assoc. Sci. Hydrol. Bull.
erosion processes and the size selectivity of eroded sediment on steep slopes using 56 (7), 1289–1299.

11

You might also like