1 Introduction

Artificial soil erosion caused by engineering measures has become more severe worldwide (Cerdà et al. 2007; Peng et al. 2014; Chen et al. 2016). Compared with the erosion process of natural soil, that of sloped farmland soil after farmland consolidation engineering is more complex (Sklenicka 2006). The new reconstructed soil which exhibits unstable behavior immediately after the implementation of farmland consolidation projects is in the restoration (RP) period (Yu et al. 2010). Thus, the soil erosion of the new reconstructed soil should be fully considered and investigated. Existing research has found that the effects of farmland consolidation engineering on the soil structural properties were significant (Wang et al. 2011). For example, the sand and aggregate disruption (PAD) was significantly reduced, and the silt, clay, and mean weight diameter of the soil aggregates (MWD) increased significantly (Ramos et al. 2007; Liu et al. 2015). The soil physical quality was much lower and most unstable 6~12 months after land consolidation (Xu et al. 2009; Curtaz et al. 2015). However, the effects of erosion on the soil structure of new reconstructed soil are rarely studied. New reconstructed soil is widespread in the hilly areas of Southwest China (Liu 2015). Particularly, purple soil corresponds to Regosols according to the FAO/UNESCO classification system (FAO/Unesco 1988), which occurs mainly in Chongqing and Sichuan of China which easily eroded and serve as the main sediment source in the Three Gorges Reservoir area (Xie et al. 2005). Acid purple soil contains more montmorillonite in which the stability of soil aggregates is poorer than other Regosols soils (Wakindiki and Ben–Hur 2002), because the parent sedimentary rock environment of the purple soil belonging to Shaxi Miao Formation is a river with relatively dry conditions (Wei et al. 2006; He et al. 2009). Previous studies have evidently shown the characteristics of the soil loss and hydrodynamic parameters of new reconstructed purple soil during the erosion process (Zuo et al. 2018). The erosion resistance of new reconstructed purple soil is weak due to its high gravel content, deteriorating physical properties, and low organic matter content (Wei et al. 1994; Wei et al. 2000). Hence, the effects of erosion on the soil structure of new reconstructed purple soil are considered to be unique.

The soil particle-size distribution (PSD) and soil aggregate stability have been demonstrated to be the key factors controlling soil erosion (Fiès and Bruand 1998). In particular, there are contradictive views on the effect of the soil PSD and aggregate stability on erosion. High silt and (or) clay contents could lead to high soil erosion (Ampontuah et al. 2006; Piccarreta et al. 2006), which is opposite of the conclusion of Wu et al. (2017). Numerous studies have demonstrated that fine particles (such as silt and fine clay) are more easily eroded by runoff than coarse materials (Starr et al. 2000; Refahi 2000; Nadeu et al. 2010). Many studies have revealed negative relationships between aggregate stability and soil erosion, for instance, in the aggregate stability indexes of the water-stable aggregate content, macroaggregates, microaggregates, and MWD (Refahi 2000; Fox et al. 2004; Thomaz 2018). However, microaggregates, which are < 0.05-mm aggregates and (or) 0.5~1- and 0.25~0.5-mm soil aggregates, were preferentially transported in the erosion process (Shen et al. 2008; Hao et al. 2019). However, most studies focused on the changes of soil PSD and aggregate stability after erosion of stable soil without further investigation on the unstable soil. Thus, the effects of erosion on the soil PSD and aggregate stability of new reconstructed purple soil should be quantified in detail to effectively carry out soil and water conservation and modeling.

The runoff, slope length, and slope position are commonly used to describe the degree of soil erosion (Meyer and Monke 1965; Liu et al. 2000; Tian et al. 2015). The influence of rainfall or runoff on the aggregate size is debatable. On the one hand, the rainfall intensity might play a negative role in the erosion process, and the detach ability of macroaggregates has been shown to increase with the increase in rainfall (Shen et al. 2008; Lu et al. 2016; Wu et al. 2017). On the other hand, increasing rainfall might cause the MWD to first decrease and then increase in the sheet flow (Hao et al. 2019). When the rainfall was greater than a certain threshold, the macroaggregates (> 0.25 mm) disintegrated seriously (Wang et al. 2014; Hao et al. 2019). Moreover, a large proportion of the fine soil aggregate components of the runoff was found when rainfall was light (López-Tarazón et al. 2010; Shi et al. 2012) and coarser particles were associated with higher stream power (Asadi et al. 2011). Many previous studies have also found that the erosion amount decreased, increased, or fluctuated upon increasing the slope length (Stomph et al. 2002; Liao et al. 2008; Kara et al. 2010). It was also found that the upslope area was erosive area, while mid-slope and lower slope areas were depositional sites. For example, Kosmas et al. (2001) found that massive amount of clay was deposited in the lower slope position. Chen et al. (2010) reported that clay and silt decreased in upslope areas, clay and silt increased in mid-slope areas, and clay and sand increased in lower slope areas. Wang et al. (2018) indicated that the macroaggregate contents at depositional sites were higher than those in eroded slope farmland areas. In contrast, Ampontuah et al. (2006) found that the clay and fine silt contents on upper slope were higher than those in lower slope samples. Therefore, the effects of the overland flow discharge, slope length, and slope position on soil erosion should be systematically studied, especially through field experiments on new reconstructed purple soil.

To bridge the knowledge gaps, this study sought to analyze the effects of erosion on the soil structure of new reconstructed purple soil in overland flow under different flow discharges, slope lengths, and slope positions. The current study specifically aimed to (i) measure the changes of the soil PSD and aggregate stability of new reconstructed purple soil in overland flow under different flow discharges, slope lengths, and slope positions and (ii) establish the relationship between the most correlative structure index and soil erosion. The results of this study could provide important parameters for predicting soil erosion through monitoring soil structural variables which can be used to control the erosion of new reconstructed slope soil in hilly areas with purple soils.

2 Material and methods

2.1 Experimental sites

This study was conducted at the national purple soil monitoring base of Southwest University, Beibei District, Chongqing (106° 26′ E, 30° 26′ N, and 230 m a.s.l.) (Fig. 1). The study site is located in the transition zone between the Qinghai-Tibet Plateau and the middle reaches of the Yangtze River. The topography of the study site is characterized by the Beibei Syncline of the parallel mountain ridge and valley region in east Sichuan. The area has a humid subtropical monsoonal climate with a mean annual temperature of 18.3 °C and an average annual precipitation of 1105 mm. The soil type is purple soil developed from the gray-brown-purple sandy mudstone of the Shaximiao Formation (J2s) of the Jurassic system. The soil properties before reconstruction are listed in Table 1.

Fig. 1
figure 1

Layout of the experiment plots at Southwest University, Chongqing

Full size image
Table 1 Soil properties before the farmland consolidation engineering at Southwest University, Chongqing
Full size table

At the experimental site, farmland with a 20° slope was selected, and the slope angle was reduced to 10° by converting the sloped land into terraces (Chongqing Land Development and Consolidation Project Construction Standard 2007). The farmland consolidation was performed according to the following steps. First, a 20-cm topsoil layer was stripped from the original slope farmland. Second, engineering measures including deep excavation, refilling, land-reshaping, and leveling were implemented. Rectangular experimental plots were established without fertilization and crops. The slope length was designed to be 5, 10, 20, 30, 40, or 50 m (Fig. 1). The width and depth of the plots were 2 m and 0.6 m, respectively. The plots were separated by a 0.2-m-wide cement ridge built on the parent material. The bottoms of the plots contained natural soil parent material. The field scouring experiments were carried out immediately after completion of the farmland consolidation. Before the field scouring experiments, approximately 15 cm of topsoil in the experimental plots was loosened (equivalent to the farmland tillage depth) to implement tillage similar to that generally used in agricultural land.

2.2 Design and measurements of the field scouring experiment

The field scouring experimental design is shown in Fig. 2. Based on the observation data collected from local runoff plots, topographical conditions, and maximum amount of runoff (Liu 2015), the simulated flow discharges on the new reconstructed purple soil were 5, 15, and 30 L/min. Eighteen field scouring experiments were conducted in this study; however, there was no runoff in the 40-m and 50-m plots at 5 L/min. A storage bucket was used to supply the water, and the overflow chute was placed horizontally at the top of the test area, such that the surface water flow remained sheet-like and uniformly distributed. The top of the overflow chute was equipped with a faucet to regulate the flow discharge. A pump and flowmeter were used to provide stable flow discharge. The plots were raked flat and allowed to stand for about a week before the field scouring experiments to prevent the influence of the surface roughness and initial soil moisture content on the experimental results. The soil was saturated for 12 h before the field scouring experiments. The experiment was started when the error between the actual and designed flow discharges was less than 5%. The flow supply was stopped until the runoff was steady. The runoff bucket was placed under the collecting tank to collect the surface runoff when surface runoff was generated. After flow supply stopped, it continued to receive the surface runoff until the end of the runoff production. Sediment samples were collected in 550-mL bottles at 1-min intervals for the first 10 min and at 3-min intervals thereafter. The sediment yield was calculated by the drying method (105 °C). We selected three slope positions (upper, middle, and lower slope) in the slope length of the 5-, 10-, 20-, 30-, 40-, and 50-m plots. Three topsoil (0~10 cm) samples for replication were collected from each of the slope positions before and after the field scouring experiments of the three flow discharges. The sampling points were randomly selected from areas eroded by runoff. Although there was no runoff in the 40-m and 50-m plots, the soil structural properties changed at different slope positions, and we also collected soil samples from these plots. The soil samples were collected for further determination of the soil PSD and soil aggregate stability. The soil properties before the field scouring experiment are shown in Table 2.

Fig. 2
figure 2

Layout of the experimental setup

Full size image
Table 2 Mean values, standard deviation, and coefficients of variation (CVs) of the soil PSD, > 0.25-mm aggregate destruction rate (PAD0.25), soil mean weight diameter (MWD), and < 0.02-mm soil aggregation degree (Y) for each plot before runoff scouring
Full size table

The erosion rate is the amount of sediment produced on the slope per unit time and can be obtained by

$$ {E}_{\mathrm{r}}=\frac{M}{t} $$
(1)

where Er is the erosion rate (g/min), M is the sediment weight in the 550-mL bottles (g), and t is the sampling time (min).

2.3 Measurement of the soil particle-size distribution and soil aggregate

In the laboratory, the soil samples were naturally dried and divided into two equal parts. The first sample was sieved through 2-mm mesh to analyze the soil PSD distribution, microaggregates, and pH, whereas the second sample was used to assess the soil aggregates. The properties of each soil sample were determined in triplicate. Soil pH was determined by potentiometry in water using a 1:5 soil-water extract. The pipette method with Na hexametaphosphate after soil organic matter (SOM) oxidation with H2O2 was used to measure the soil PSD (Zheng 2013). Soil microaggregates were determined using the pipette method without the removal of the SOM. According to the soil texture classification system of the USDA, the soil particles were divided into sand (2~0.05 mm), silt (0.05~0.002 mm), and clay (< 0.002 mm). The size distribution of the dry-stable aggregates was determined by the dry sieving method. Soil samples were placed on a stack of sieves (10, 7, 5, 3, 2, 1, 0.5, and 0.25 mm) and vibrated for 10 min (SZH, Nanjing Soil Instrument Factory, China). The water-stable aggregates with a diameter of > 0.25 mm were analyzed by the conventional wet sieving method (TPF-100, Nanjing Soil Instrument Factory, China). Stacked sieves (5, 2, 1, 0.5, and 0.25 mm) were put into the water and moved up and down for 30 min at a frequency of 30 times/min (Oades and Waters 1991).

The soil aggregate size of 0.25 mm represents the boundary between macroaggregates and microaggregates (Six et al. 2000). The soil aggregate stability was indicated by the mean weight diameter of the soil aggregates (MWD), the > 0.25 mm percentage of aggregate disruption (PAD0.25), and the degree of aggregation (Y) (Zhang et al. 2014). These values were calculated with Eqs. (2) to (4). The index Δ was considered to be the change in the soil structure affected by erosion in overland flow and was calculated as follows: Δ = (the indicator value after scouring − the indicator value before scouring) ∕ the indicator value before scouring.

$$ \mathrm{MWD}={\sum}_{i=1}^6{x}_i{w}_i $$
(2)

Here, xi is the mean diameter of each size class (< 0.25 mm, 0.25~0.5 mm, 0.5~1 mm, 1~2 mm, 2~5 mm, and > 5 mm), and wi is the weight fraction of aggregates in size class i.

$$ {\mathrm{PAD}}_{0.25}=\frac{\left({D}_{0.25}-{W}_{0.25}\right)}{D_{0.25}}\times 100\% $$
(3)

Here, D0.25 is the > 0.25-mm dry-sieved aggregate content, and W0.25 is the > 0.25-mm water-stable aggregate content.

$$ Y=\frac{a^{\prime }-{b}^{\prime }}{b^{\prime }}\times 100\% $$
(4)

Here, a′ is the > 0.02-mm water-stable aggregate content, and b′ is the > 0.02-mm mechanical composition.

3 Results

3.1 Characteristics of the soil particle-size distribution, soil aggregate stability, and erosion rates after farmland consolidation

The clay content was 16.14~22.17% and the silt and sand contents were 39.04~41.27% and 37.54~44.82%, respectively, before farmland consolidation (Table 1). The soil PSDs of the new reconstructed purple soil showed that the soil texture was clay loam with a high content of silt particles (36.69~43.05%) and that the silt content was maximal in the 10-m plot before runoff scouring (Table 2). Before runoff scouring, the MWD, PAD0.25, and Y of the new reconstructed purple soil were 99.35~159.95, 95.48~96.98%, and 11.51~29.66%, respectively, and the minimal PAD0.25 was 95.48% at 10 m (Table 3). The range of the average erosion rate was 1.11 (5 m, 5 L/min)~239.09 g/min (10 m, 30 L/min), as shown in Table 3. The partial correlation analysis revealed that the effects of the erosion rate on the flow discharge and slope length were significant (r = 0.585, p < 0.05 and r = 0.561, p < 0.05, respectively). The erosion rate increased with the increase in the flow discharge. The change in the erosion rate was not obvious with the increase in the slope length; however, it was the largest in the 10-m plot.

Table 3 The average erosion rates of new reconstructed purple soil under different slope lengths and flow discharges
Full size table

3.2 Effects of erosion on the soil structure in overland flow under different flow discharges, slope lengths, and slope positions

The ANOVA revealed that the effect of the flow discharge on the PAD0.25 was highly significant (p < 0.01) (Table 4) and that the slope length was significantly correlated with the MWD (p < 0.05). Other measured effects were not significant, but some trends in the changes after runoff souring were still observed (Fig. 3). The clay, silt, and sand variables before and after scouring are shown in Fig. 4 and Table 5. The MWD, PAD0.25, and Y variables before and after scouring are shown in Fig. 4 and Table 5. The response of the soil particle-size distribution (PSD) and soil aggregate stability to erosion in new reconstructed purple soil under different flow erosion rates was different. At 30 L/min, the maximal average reduction rate of silt was 2.11%, and sand particles showed an increase of 1.27%. The average ΔMWD and ΔPAD0.25 were − 11.46% and 0.70%. The maximal average increase rate of Y was 27.64%. At 15 L/min, clay and silt were eroded simultaneously at average reduction rates of 0.80% and 1.00%, respectively, while the sand increased by 2.02%. The average reduction rate of the MWD was 20.23%. The average ΔY was 9.52%. At 5 L/min, the sand content increased by 1.11%. Clay and sand were removed at 1.04% and 0.11% on average. The maximal average increasing rate of the PAD0.25 was 0.86%. In contrast, Y decreased by only 5.61% at 5 L/min. Overall, the decreasing rate of silt increased with the increase in the flow discharge. The content of silt and the soil aggregate stability both decreased, but sand content increased under the three flow discharges.

Table 4 Effects of flow discharge, slope length, and slope position on soil structure and their contributions based on ANOVA
Full size table
Fig. 3
figure 3

Soil PSD variation in the topsoil before and after runoff scouring

Full size image
Fig. 4
figure 4

Comparison of the soil aggregate stability index in the topsoil before and after runoff scouring

Full size image
Table 5 The average change rates of soil PSD and aggregate stability under different flow discharges, slope lengths, and slope positions
Full size table

There were no obvious changes in the soil structure of the new reconstructed purple soil with the increase in the slope length under erosion. However, the negative effect of erosion on the soil structure in the 10-m plot was serious. The maximal average reduction rate of silt (7.72%) and the maximal average increasing rate of sand (16.70%) both occurred in the 10-m plot. Meanwhile, the maximal average reduction rate of the MWD (25.18%) and the maximal average increasing rate of the PAD0.25 (1.86%) both occurred in the 10-m plot. The structure of the new reconstructed purple soil under the influence of the slope position on erosion was obviously different. The maximal average reduction rate of the MWD was 14.32% at the upslope position and the ΔPAD0.25 was 0.63%. Clay and sand particles showed an increase of 0.15% and 0.09%. The average silt and Y value decreased by 0.19% and 6.11%. At the middle slope position, the average clay and silt contents increased by 3.10% and 6.39%, respectively, while the average sand content decreased by 10.06%. The average reduction rate of the MWD was 12.98%, and the increasing rate of the PAD0.25 was 0.69%. The maximal average increase in Y was 13.00%. At the downslope position, changes to the Y value and clay particle content showed an average increase of 4.58% and 5.11%. The silt and sand contents continued to decrease by 0.52% and 3.75%. The minimal average reduction rate of the MWD was 7.87% and the maximal average increasing rate of the PAD0.25 was 0.72%. Overall, the reduction rate of the MWD decreased gradually from the position of the upper, middle, and lower slopes and the trend of the PAD0.25 was opposite. The upslope area showed the destruction of aggregates. The middle slope was the main sedimentary area of silt and the area of microaggregate destruction. Microaggregates and clay particles were deposited on the lower slope areas.

3.3 The correlation between the erosion rate and soil structural variables

With the average erosion rate as a reference series, and the indicators of the topsoil structural variables as a comparative series, gray correlation analysis was performed (Deng 1987). The results indicated that the degree of the response of changes in the topsoil structural parameters to the erosion rate was ranked in the following order: γMWD (0.50) > γclay (0.47) > γsand (0.41) > γsilt (0.40) > γQ (0.37) > γPAD (0.26) > γL (0.21) > γY (0.16). The response of the soil structural parameters to the erosion rate was correlated in the new reconstructed purple soil, with the ΔMWD having the most significant effect on the erosion rate and Δclay having the second-most significant effect. As shown in Fig. 5, based on the ΔMWD and erosion rate, the formulas for predicting the erosion rate as established by exponential correlation were R2 = 0.8187 (30 L/min) and R2 = 0.6483 (5 L/min).

Fig. 5
figure 5

Relationship between change rate of soil MWD (ΔMWD) and erosion rate on new reconstructed purple soil

Full size image

4 Discussion

4.1 Influence of the overland flow discharge on the soil structure of erosion under new reconstructed purple soil

The flow discharge rate influenced the erosion of new reconstructed purple soil in the hilly area (p < 0.05) in this study. The rainfall or runoff amount and (or) intensity are commonly used to describe the degree of soil erosion (Meyer and Monke 1965; Peng et al. 2014; Prosdocimi et al. 2016). The average erosion rate of new reconstructed purple soil increased with increasing flow discharge, which was consistent with the results of most studies (Hamed et al. 2002; Kang et al. 2016). First, the upper water inflow increased with the increase in the flow discharge, and the runoff of the slope soil increased per unit time. Therefore, the high surface runoff velocity increased the kinetic energy of erosion and the crushing capacity of soil particles in the new reconstructed purple soil.

Under different flow discharges, silt was preferentially eroded and soil aggregates were destroyed at 30 L/min, clay and silt were eroded simultaneously and the soil aggregate stability was damaged most seriously at 15 L/min, and clay loss was maximal and the macroaggregate were destroyed most severely at 5 L/min in our study. The microaggregates increased at 30 and 15 L/min, and the changes of the clay content were consistent with those of the microaggregates because clay is an important component of microaggregates (Bronick and Lal 2005) (Table 5). Previous studies showed that the destruction of soil macroaggregates was positively correlated with rainfall or runoff (Shen et al. 2008; Lu et al. 2016), which was not corroborated by our results. This difference may be explained by the influence of erosion on the soil PSD having priority under high flow discharge with the original PSDs of new reconstructed purple soil. Our study showed that the sand content increased under all three flow discharges, with the maximal increasing rate at 15 L/min (Fig. 3, Table 5). It was further shown that the destruction of aggregates was the source of sand particles, because the average maximal reduction rate of the MWD occurred at 15 L/min. Previous studies have also shown that PSDs after scouring were significantly coarser than those of the original soil (Chen et al. 2016; Ding and Huang 2017). This result can be explained by fine material being more easily transported than coarse material (Starr et al. 2000; Nadeu et al. 2010). We found that silt particles were the primary type of particles transported in the erosion of new reconstructed purple soil, as inferred from the increase in the decreasing rate of silt with the increase in the flow discharge matching that of the erosion rate. Additionally, the variation trends of silt and the PAD0.25 with an increase in the flow discharge were contradictory (Table 5). This result showed that the silt loss was not from macroaggregates and was mainly from the soil matrix of the new reconstructed purple soil, which was different from the thought that macroaggregate breakdown is responsible for the production of fine materials (Le Bissonnais 2005; Shen et al. 2008). This is more likely attributable to the PSDs before scouring, in which the silt particle content was 36.69~43.05% in the new reconstructed purple soil (Table 2). This result is consistent with that of the high silt content having increased the soil erodibility due to its poor aggregation (Meyer et al. 1992). Refahi (2000) found that soils with 40~60% silt contents were the most vulnerable to erosion. Furthermore, purple soil particles develop from the fast weathering of siltstone, mudstone, and sandstone. In particular, the parent sedimentary rock environment of the Shaxi Miao Formation is a river with relatively dry conditions, leading to low organic matter and clay content (Wei et al. 2006; He et al. 2009). Additionally, Boonamnuayvitaya et al. (2004) indicated that particles less than 0.02 mm in diameter were particularly important for chemical transport, due to their large surface areas.

There was an exponential relationship between the ΔMWD and erosion rate when the flow discharge was 30 or 5 L/min as shown in Fig. 5, as the MWD increased the soil erosion resistance (Refahi 2000). This finding indicated that the soil aggregate stability properties of the new reconstructed purple soil were the main factors affecting the erosion rate under simulated high and low overland flow conditions. The large amount of water made the sediment and flow pattern more obvious; therefore, a high flow discharge increased the erosion intensity. This result agrees with the findings of Barthès and Roose (2002) and Fox et al. (2004), who clarified the effects of the water-stable aggregate and macroaggregate contents on the erosion amount by a negative correlation. In application of the models, some erosion models have begun to use the data of the soil structural characteristics as output or input parameters (Gholami et al. 2018). K is important for efficiently quantifying the soil erodibility efficiently in many models, having an obvious relationship with the soil particle-size distribution (PSD) and soil aggregate stability (Oztas and Fayetorbay 2003; Li et al. 2015).

4.2 Influence of the slope length and slope position on the soil structure of erosion under new reconstructed purple soil

The slope length is one of the key variables in empirical model of soil loss (Liu et al. 2000). This study found that changes of the soil structure were not obvious with the increase in the slope length; however, the damage to the soil PSD and aggregate stability in the 10-m plot was the worst (Table 5). Because the maximal average reduction rate of silt (7.72%) and the maximal average increasing rate of sand (16.70%) both occurred in the 10-m plot, these results showed that it was vulnerable to erosion. Additionally, the maximal average reduction rate of the MWD (25.18%) and the maximal average increasing rate of the PAD0.25 (1.86%) both occurred in the 10-m plot due to its poor soil aggregate stability (Table 5). Meanwhile, the maximum values of the average erosion rates occurred in the 10-m plot when the flow rate was 30 L/min and 15 L/min (Table 3). The soil properties of the new reconstructed purple soil with a slope length of 10 m before souring showed poor aggregate stability and weak erosion resistance, such as the maximal silt content and the minimal PAD0.25 both being in the 10-m plot before runoff scouring (Table 2). Additionally, the runoff shear stress, which causes soil particle separation and sediment transport, was high in the 10-m plot at our experimental site (Zuo et al. 2018).

Our study showed that the destruction of the soil aggregate stability decreased gradually following the order of the upper to the middle and lower slopes, due to ΔMWD decreased gradually with the position of upper, middle, and lower slopes (Table 5). The upslope area showed intense destruction of aggregates. The mid-slope area was the main sedimentary area of silt and the area of microaggregate destruction. The lower slope areas were comprehensive depositional sites, in which microaggregates and clay particles were mainly deposited, and the reduction rates of the silt and sand contents in the lower slope areas were smaller than those in the middle slope areas (Fig. 3, Table 5). These results were different from those obtained for disturbed soil accumulation, in which clay and silt were transported mainly from the middle slope positions (Peng et al. 2014), but they were consistent with those obtained for other effects of human activities such as intense tillage (Chen et al. 2010). Generally, the runoff from the upper slope position increased in flow energy and velocity at the middle slope positions (Wang et al. 2004). Further, the middle slope position was the most complex location, where the process of deep excavation and filling coexisted with the process of farmland consolidation engineering. Soil transport and detachment coexisted at the middle slope position. Other studies have noted that clay particles were transported and deposited in large quantities to downhill positions (Chen et al. 2010). These results revealed that the soil aggregates were destroyed by erosion at all slope positions of the new reconstructed purple soil, with a variation tendency similar to those in the previous findings (Chen et al. 2010; Wei et al. 2017). However, the amplitude of change in this study was smaller than those in the others studies. Chen et al. (2010) reported that the PAD0.25 increased maximally at the upper slope position in purple soil runoff plots (grade = 25°). This difference may be attributable to the gravity component of the water gravitational potential energy decreasing with a decrease in the slope gradient. The water gravitational potential energy of the 25° steep slope was larger than that of the 10° slope. The large slope gradient enhanced the runoff detachment and transport capacity and increased the soil aggregate loss (Poesen 1984). Furthermore, this study employed a field scouring experiment without the impact of raindrops and splash corrosion. In addition, the depth of the land and mechanical compaction can lead to changes in the structural profile of soil during the process of land consolidation. Although surface soil stripping and backfilling will greatly reduce this change, it is very difficult to produce uniform topsoil for backfilling (Wang et al. 2011). According to the above results of this research, soil and water conservation measures should be implemented on the upper slopes of new reconstructed purple soil with a slope length of 10 m which can effectively and economically control soil erosion. Soil PSD and MWD variables could be used as parameters for prediction of soil erosion on new reconstructed purple soil.

Although this experiment is only a case study, it is the first to explore the influence of the erosion on the soil structure of new reconstructed purple soil. Due to the limitations of experimental conditions, there are still some shortcomings in this experiment. The study is limited in that it involves only a single test of every experimental treatment. However, the analysis of this study mainly focuses on the effects of the flow discharge, slope length, and slope position on the soil structure in overland flow. Multiple sets of experimental data are collected under each influencing factor to ensure the accuracy of the results as much as possible. In addition, new reconstructed purple soil in the field is sensitive to natural rainfall. After experiencing the erosion from multiple rains over a period of time, new reconstructed purple soil will gradually change from the unsteady state to the stable state. To ensure the original state of the new reconstructed purple soil as much as possible, our experiment was completed within a short amount of time to reduce the impact of external environmental factors on the sample plot. Future work should focus, in particular, on the effects of artificial rainfall simulation on the soil structure and the fertility of the new reconstructed purple soil. Additionally, changes to the characteristics of the soil structure changing over time during erosion progression are essential for drawing comprehensive conclusions.

5 Conclusions

Erosion had significant and different effects on the soil PSD and aggregate stability of new reconstructed purple soil in overland flow under different flow discharges, slope lengths, and slope positions. (1) Silt was the primary particle transported during the erosion, and the decreasing rate of silt increased with the increase in the flow discharge. Silt was eroded most at 30 L/min, clay and silt were eroded simultaneously, and the soil aggregate stability was damaged most seriously at 15 L/min; clay loss was maximal and the macroaggregate were destroyed most severely at 5 L/min. The average erosion rates of new reconstructed purple soil increased with increasing flow discharge. The relationship between the erosion rate and ΔMWD was significantly exponential under high and low overland flow discharges. (2) The damage to the soil PSD and aggregates stability of the 10-m plot was the worst. The silt and MWD maximally decreased by 7.72% and 1.86%, and the sand and PAD0.25 maximally increased by 16.70% and 25.18%, which all occurred in the 10-m plot. (3) The destruction of the soil aggregate stability decreased gradually following the order of the upper to the middle and lower slopes. The upslope showed intense destruction of aggregates. The middle slope was the main sedimentary area of silt and the area of microaggregate destruction. The lower slope areas were comprehensive depositional sites, in which microaggregates and clay particles were mainly deposited, and the reduction rates of the silt and sand content were smaller than those in the middle slope. Hence, it is better to implement soil and water conservation measures on the upper slope of new reconstructed purple soil with a slope length of 10 m which can effectively and economically control soil erosion. The soil PSD and MWD variables could be used as parameters for the prediction of soil erosion in new reconstructed purple soil.