1. Introduction
The Loess Plateau is one of the regions characterized by the most serious soil erosion in China. The annual precipitation in this region is less than in surrounding areas, the ecological environment is poor, and the vegetation recovery is diminished [
1,
2,
3,
4]. Due to erosion, a greater amount of sediment particles is transported to the Yellow River via runoff and deposited in the lower reaches, resulting in the elevation of the riverbed, and posing a serious threat to the ecological security of the lower reaches [
5]. Therefore, the prevention and control of soil erosion has become the focus of the response to these global environmental issues [
6,
7,
8]. Due to the variation in the physical characteristics, infiltration [
9], water conductivity, and water holding capacity between the surface sand layer and the yellow soil layer, a sand interface is formed, which in turn forms the typical dual structure of sand [
10,
11,
12,
13].
Water infiltration is an important component of the hydrological cycle [
14,
15,
16,
17], as it directly determines the generation time and size of slope runoff, and also affects soil moisture at different depths [
18,
19]. During this process, rainfall leads to changes in the soil infiltration capacity on the slope due to sand coating, which further changes the runoff on the slope [
20]. In this way, the runoff depth increases, resulting in changes in hydraulic characteristics [
21]. Previous studies have investigated this infiltration process, the runoff and sediment yield process of sand-covered slopes, and their relationship, as well as the influence of sand grain size composition on the erosion process [
4,
10,
11,
22]. In a field rainfall experiment, Zhang et al. [
23] found that the runoff process of sandy slopes was obviously different than that of loess slopes. Here, the runoff of sandy slopes decreased, but the sediment content in the runoff increased. Similarly, Wu et al. [
24] qualitatively described the interfacial flow of sandy soil on sandy slopes through field investigation. Additionally, through an indoor simulation study, Tang and Su [
22,
25] found that the initial runoff time of sandy slopes increased, and, with increasing sand thickness, under different treatments, the cumulative sediment yield increased with the increase in runoff.
Slope runoff velocity is one of the most important soil moisture parameters, and has great influence on soil erosion. Slope runoff and sediment movement are closely related to hydraulic parameters [
26]. The main factors affecting the hydraulic characteristics of slope are slope sand covering, soil freezing, and so on [
27,
28,
29]. Previous work has observed that the flow pattern of water has a great influence on the erosion of sandy slopes. Similarly, it has been determined that under different treatment conditions, the two parameters describing the runoff process are runoff velocity and the hydraulic parameters [
27,
30].
Therefore, through an indoor simulated rainfall experiment, taking the loess and sand-covered slopes in the east willow ditch in northern Shaanxi as the research object, the infiltration process, runoff characteristics, and influencing factors of the sand-covered slope were analyzed, the variation process of the soil moisture parameters was studied, and the relationship between erosion sediment yield parameters and soil moisture parameters was explored. Specifically, slope soil moisture content, time units for different parameters of runoff and sediment yield, slope runoff velocity and water depth, combined with the spatial and temporal variations of soil moisture content, were obtained via the design of different rainfall densities.
2. Materials and Methods
2.1. Test Material
Simulated rainfall experiments were conducted in the State Key Laboratory of Eco-Hydraulic Engineering at Xi’an University of Technology in China. In this test, a side jet rainfall simulation device was used, with uniformity >90%. The simulated rainfall test used a wooden soil bin that was 2 m long, 0.75 m wide, and 0.60 m high. The wood was 3 cm thick, which meant that the soil could be kept warm and that a one-dimensional thaw occurred in the vertical direction of the soil. The lower end of the soil bin was connected to a collecting tank, which was used to collect runoff and sediment samples (
Figure 1 and
Figure 2). Soil moisture content was measured using the CR1000 data acquisition device from Campbell Company in the United States. Measurements of soil moisture content were obtained using a CS616 soil moisture sensor at a frequency of 1 measurement/min. Five water probes were positioned along the vertical direction at depths of 3 cm, 6 cm, 9 cm, 14 cm and 22 cm from the soil surface, respectively.
The experimental soil was selected from an alfalfa field in Wangmaogou, Suide, Northern Shaanxi Province. The surface soil was 20–30 cm in depth, and the sandy soil was characteristic of the aeolian sand of the Dongliugou watershed. After the soil samples were returned to the laboratory, debris such as grass roots and stones were removed before being passed through 10 mm (soil) and 0.8 mm (sand) sieves for pretreatment. Prior to filling the soil trough, the wooden soil trough was first soaked with flowers to allow for better bonding with the soil. In addition, a layer of fine sand was laid at the bottom of the soil groove, and the bottom of the soil groove was drilled so that the water can penetrate normally after reaching the deep layer. The soil bulk density was controlled at about 1.3 g/cm
3, and the soil water content was about 15%. The thickness of the soil was 40 cm, and the method of layered filling and artificial compaction was used to fill the soil. The thickness of each layer was about 5–8 cm. After each layer was filled, the surface of the soil layer was haired to closely combine the two layers of soil. After filling the soil tank, the soil surface was covered in 2 cm of sand according to the experimental design. The adhesion to the surface soil was noted during the sand-covering, so the method of burring and sand-covering followed by compaction was adopted. Finally, in order to ensure that the test conditions of each test were consistent, the slope was leveled every time, and non-invasive rainfall with rainfall density of 30 mm/h was applied before the slope rainfall, after which the test soil tank was covered with plastic cloth for a waiting period of 24 h before testing. The characteristics of soil particles were measured using a Malvin 2000 (Malvern Worcs, WR141XZ, UK). The characteristics of the soil particles are shown in
Table 1.
2.2. Experimental Design
This experiment follows a simulated rainfall test design. According to previous literature and the investigation of the thickness of a sand layer deposited in the water–wind erosion crisscross region, the thickness of sand cover was determined to be 2 cm. Through the study of a simulated rainfall experiment, Zhou et al. [
28] found that the standard of the loess plateau erosive rainstorm intensity was 10.50~234.84 mm/h, so the rainfall densities in this experimental design were set to 1.0, 1.5, 2.0 and 2.5 mm/min. The treatment slope surface was 2 cm sand-covered loess, and the control slope was loess slope. The slope of the test soil groove was fixed at 12°. Intermittent rainfall was applied 3 times per treatment, where each rainfall lasted for 60 min, and each rainfall interval was 24 h, which was the interval when using a rain shelter cover soil trough. The test was repeated 2 times under each condition, for a total of 4 × 2 × 3 × 2 = 48 field tests. The design scheme for the test is shown in
Table 2.
2.3. Measurement and Calculation of Indicators
- (1)
Determination of Runoff Sediment Index
Before the test, the slope surface was divided into four 0.5-m-long observation sections from bottom to top. Once the rainfall began, a 5000 mL measuring tube was placed at the runoff and sediment outlet of the soil trough. As the runoff began on the slope, the measuring cylinder was used to collect the runoff sediment samples every minute, and conical bottles were used to collect the runoff sediment samples. Each conical bottle was then left to rest for about 2 h to allow for the sediment to settle, after which the supernatant was carefully removed. The remaining runoff sediment samples were then poured into an iron box with known weight, and placed in an oven to be dried at 105 °C prior to weighing. The sediment weight in the conical bottle was then obtained. Next, the sediment weight in the measuring cylinder was calculated by the replacement method [
31]. The sum of the two measurements is the total sediment yield in this time, and the sum of the runoff in the measuring cylinder and the runoff in the conical bottle is the total runoff in this time.
- (2)
Calculation of runoff velocity and soil moisture parameters
During the experiment, a potassium permanganate staining method was used to determine the surface runoff velocity,
Vs, of the slope runoff. The average runoff velocity of slope runoff was calculated as follows:
where
V is the average velocity of runoff and
β is the correction coefficient of runoff velocity, and is taken as 0.75 in this study [
32].
Since the slope flow in the experiment is a thin-layer flow, it is difficult to measure the runoff depth
h. For this reason, previous researchers have adopted the assumption that the slope flow is uniformly distributed. The method for calculating the runoff depth in this study is as follows:
where
h is runoff depth, m;
Q is the runoff during a period of
T, m
3;
V is slope average velocity, m/s;
B is water width, m; and
T is the duration, s. Runoff depth in rills is measured directly using the ruler method.
The shear force of runoff can peel soil particles from their original position by damaging the original structure of the soil and removing them from the slope with the flow. In practical research, the movement form of slope flow is simplified as a one-dimensional uniform flow. The method for calculating runoff shear stress is as follows:
where
τ is the runoff shear stress on the slope, Pa;
ρ is the density of rain water, 1000 kg/m
3;
g is the gravity acceleration, which is 9.8 m/s
2;
R is the hydraulic radius, where the hydraulic radius of the thin-layer flow is equivalent to its runoff depth, and the runoff depth of the rill is measured by a ruler, m; and
S is the hydraulic gradient, which is simplified as the sinusoidal value of the soil groove gradient, namely
S = sin θ, and θ is the soil groove gradient of 12°.
Runoff power refers to the change rate of water potential energy with time per unit of area. This study uses the following method to calculate runoff power:
where
ω is the runoff power of the slope, N/(m·s), and the other letters have the same meaning as denoted previously.
2.4. Data Processing and Analysis
In each rainfall experiment, the following data were collected: runoff generation time under two kinds of slopes, runoff velocity of slopes and rills at different rainfall times, soil moisture content, ditch width, ditch depth, and runoff per time unit.
Pearson correlation analysis was used to analyze the correlation of the collected data. Statistical and regression analyses were performed in SPSS 22.0 (Stanford University, Stanford, CA, USA) using Origin 2017 (Originlab, Northampton, MA, USA), 3ds Max (Autodesk Corporation, San Rafael, CA, USA), and Excel 2010 (Microsoft Corporation, Redmond, WA, USA).
4. Discussion
4.1. Sand-Covered Slope
In this study, the infiltration velocity of the soil was observed to decrease with increased rainfall. On one hand, the soil moisture content increases with increasing rainfall duration, which leads to a decrease in soil infiltration capacity [
33,
34]. On the other hand, the soil surface may gradually form a crust during rainfall, further preventing the infiltration of soil moisture [
35,
36,
37]. There are two main differences in the infiltration law between the sand-covered and loess slopes, namely, the initial infiltration velocity and the change rate of the infiltration velocity. During the first 20 min of rainfall, the sand-covered slope showed a greater soil infiltration velocity than that of the loess slope. However, it was observed that the rate of change of the infiltration velocity of the loess slope was greater than that of the sand-covered slope, so the time required to achieve stable infiltration of the loess slope is likely shorter. This is because the porosity of the sand layer on the surface is high, and the water is more likely to leak in the vertical direction. In addition, because the sand layer stores some rainfall, the amount of rain that seeps into the surface of the loess soil layer is lower than the amount of the actual rainfall, so the rainfall density at the initial stage does not exceed the infiltration capacity of the soil. Therefore, the infiltration velocity of the sand-covered slope is greater than that of the loess slope during this period [
13,
38,
39].
With the same duration of rainfall, the soil moisture reaches the deep soil earlier of the sand-covered slope than of the loess slope, with greater infiltration. This is likely because the sand layer of the sand-covered slope surface weakens the energy and the size of raindrops. In addition, soil crust cannot be generated under the sand layer conditions, resulting in increased soil infiltration capacity. The results of this study indicate that before steady infiltration is achieved, the water infiltration velocity of the sand-covered slope will always be greater than that of the loess slope, so the infiltration amount will also be greater. In the process of infiltration, the soil moisture content of the shallow soil increases first. By comparing the runoff time, it was found that the soil moisture content at a depth of 6 cm reached a certain value of the sand-covered slope, at which point runoff began, indicating that the shallow soil was nearly saturated. After the runoff generation time, the water can be divided into two categories. One part generates surface runoff, while the other part is used in the process of infiltration. At this point, the infiltration amount of the sand-covered slope was still greater than that of the loess slope, resulting in the increase in water content in the deep soil occurring earlier of the sand-covered slope than of the loess slope. Following rainfall, the soil moisture content at different depths was generally consistent of the sand-covered slope. In addition, it was found from the water distribution process that the soil moisture content at a depth of 22 cm remained the highest after rainfall. In the two subsequent rainfall experiments, the soil moisture content at a depth of 22 cm remained the highest, and the remaining water content curves showed a decreasing trend with increasing soil depth. Finally, the soil moisture content at different depths were basically the same between the slopes.
4.2. Soil Moisture Changes
Compared to the loess slope, the runoff velocity of the sand-covered slope fluctuated significantly with time, mainly as a result of the characteristics of sediment production. Due to the special physical properties of sand, the sand–soil dual structure becomes fragile, and the slope is therefore more prone to erosion and collapse under the action of hydraulic effects and self-gravity [
10,
11,
12,
13]. In this way, flow channels will be blocked at a given point, and subsequently washed away; therefore, runoff velocity shows great fluctuation. Of the loess slope, the runoff velocity did not fluctuate greatly with the duration of rainfall, indicating that the flow characteristics were stable throughout the entire rainfall process. The runoff velocity increased with increasing rainfall density, which is likely because increased rainfall density leads to increased water depth and hydraulic radius in the rain-affected area, whereby the resistance coefficient is decreased, and the runoff velocity increases with increasing rainfall density [
40,
41].
The water infiltration of the sand-covered slope was greater than that of the loess slope. Therefore, the total runoff of the sand-covered slope was observed to be less than that of the loess slope per time unit, resulting in the thin-layer flow of the sand-covered slope having a shallow water depth. Similarly, the runoff shear stress was lower than that of the loess slope. Under the same rainfall density, the runoff shear stress of the loess slope was greater than that of the sand-covered slope, but the runoff shear stress of the sand-covered slope increased more rapidly with increasing rainfall density, that is, the slope in the linear relationship was steeper. At a rainfall density of 0.5 mm/min, the runoff shear stress of the sand-covered slope increased at a rate of about 1.5 times that of the loess slope.
The runoff power [
42,
43] also suggests that the runoff shear stress of the sand-covered slope was greater than that of the loess slope. The equation for estimating runoff power highlights its close relationship with runoff shear stress and velocity. In this study, the runoff shear stress on the sediment-covered slope was greater than that of the loess slope, and there was no significant difference in the average velocity of the two slopes. Therefore, it was determined that the runoff power of the sand-covered slope was lower than that of the loess slope. Under the same rainfall density, the runoff power of the loess slope was generally greater than that of the sand-covered slope. However, the runoff power of the sand-covered slope increased slightly with increasing rainfall density. At a rainfall density of 0.5 mm/min, the increase of runoff power of the sand-covered slope was about 1.13 times that of the loess slope.
5. Conclusions
On the basis of our observation of the processes of infiltration, flow generation, water flow characteristics, and the spatial distribution erosion during a designed rainfall test, and analyzing the infiltration, flow generation characteristics, water content change characteristics, soil moisture parameter change characteristics, and changes in the spatial patterns of erosion and sediment yield, this study draws the following conclusions:
Under different rainfall densities, the initial runoff generation time of the sediment-covered slope was 1~12 min longer than that of the loess slope; the initial soil infiltration rate of the sediment-covered slope was about 1.23 times that of loess slope; and the time taken to achieve stable infiltration of the loess slope was shorter than that of the sediment-covered slope.
Under different rainfall intensities, the rising time of the water content curve of the sand-covered slope was shorter than that of the loess slope. Within the same duration of rainfall, the vertical infiltration performance of soil water of the sand-covered slope was higher than that of the loess slope.
When the rainfall intensity on the slope increased by 0.5 mm/min, the increase in the value of runoff shear force on the sediment-covered slope was about 1.5 times that of the loess slope, and the runoff power was about 1.13 times that of the loess slope.