Bedrock Fragment Induced by Intensive Tillage Effect on Hydrological Properties and Erosion Processes under Different Rainfall Patterns

Abstract

To investigate the influence of bedrock fragmentation by intensive tillage on the hydrological characteristics and soil erosion processes on slopes, two experimental treatments (soil–bedrock mixtures, WB, and pure soil, CK) in steel tanks were subjected to simulated rainfall under five rainfall patterns (constant, increasing, decreasing, decreasing–increasing, and increasing–decreasing) with the same total rainfall of 90 mm. For each rainfall event, runoff and sediment concentration were sampled at regular intervals. The flow velocity (v), effective/critical shear stress (τ/τ_c), Darcy–Weisbach resistance coefficient (f), unit stream power (p), and soil erodibility factor (Kr) were calculated to analyze the differences in hydrodynamic characteristics between the WB and CK. Our experimental findings show that significant differences in runoff volume and sediment yield were observed among different rainfall patterns and stages. Bedrock fragmenting significantly promoted runoff and sediment production under different rainfall patterns, with runoff volume and sediment yield increasing by averages of 59.29% and 71.62%, respectively. An increasing trend in average contribution rate of bedrock to runoff volume and sediment yield was observed across three distinct intensities: 6.37% and 4.61% for 30 mm h⁻¹, 12.53% and 7.53% for 90 mm h⁻¹, as well as 14.79% and 36.98% for 150 mm h⁻¹, respectively. The v and p values under various rainfall patterns exhibited an increasing trend from the upper to the bottom slope positions, whereas the f and τ values showed an opposite trend, regardless of the WB and CK. Compared with the CK, the v, f, and p values for the WB increased by 23.34% to 48.94%, 1.59% to 53.16%, and 3.86% to 27.86%, respectively, whereas the τ value decreased by 1.52% to 22.19% for varying-intensity rainfall patterns. Among the variable rainfall patterns, the WB significantly increased sediment yield and also had a promoting effect on runoff generation. However, the WB displayed better erosion resistance compared to the CK under constant rainfall patterns. Therefore, varied-intensity patterns had a profound impact on bedrock-induced runoff and sediment transport processes.

1. Introduction

Soil erosion is a global-scale ecological and environmental issue, and human-induced soil erosion exacerbates the loss of soil resources and a decline in land productivity. This problem is particularly severe in agricultural production heavily influenced by human activities [1]. Tillage operation is one of the essential human activities in agriculture, and soil erosion caused by tillage is unavoidable, especially in areas with significant topographical variations. The interactions between tillage operations and gravity result in notable displacements of cultivated soil in mountain landscapes [2].

Purple soil is an important agricultural soil in subtropical regions of China, primarily concentrated in the upper and middle reaches of the Yangtze River, with widespread distribution in Sichuan Province (accounting for 68.7% of the total arable land in Sichuan Province) and playing a critical role in the national economy [3]. Due to the presence of poorly cemented calcium-rich and clay rocks in the purple soil parent material, physical weathering of rocks is relatively strong [4]. Therefore, purple soil inherits characteristics from its parent rocks, such as high coarse particle content, low organic matter content, few aggregates, and strong dispersion [4,5]. Moreover, purple soil in in Sichuan Province is predominantly hilly, densely populated, with a long history of cultivation, intensive land use, significant human disturbance, high-intensity rainfall, favorable erosion conditions, and long-term intense cultivation. These factors have led to a continuous thinning of the topsoil on sloping farmlands, eventually exposing the bedrocks, resulting in a significant decline in land productivity [6,7]. Surveys indicate that tillage erosion in this region has caused crop yields at the top of slopes (e.g., wheat, corn, and sweet potatoes) to be only 50% or less compared to the bottom of slopes [8]. This severely limits the sustainable development of local agricultural systems.

To counteract the decline in soil productivity caused by tillage erosion, farmers often mix weathered parent material into the soil to maintain soil depth, leading to an increase in the rock fragment in purple soil [9]. Studies have shown that purple soil in the Three Gorges Reservoir area exhibits shallow soil and gravelization characteristics, with gravel contents ranging from 14.2 to 34.4% for particles of 1–3 mm and 51.2 to 74.1% for particles larger than 3 mm [10,11]. It is evident that thin soil layers and high gravel contents are essential characteristics of purple soil. Moreover, the terrain in the region of purple soil is generally hilly and low to mid mountainous, with steep slopes on cultivated lands and predominantly downhill tillage. These soils have strong erodibility, high rates of soil and water loss, and loose and easily fragmented bedrock material, which, especially in the case of purple bedrock, develops numerous fractures. When the cultivated depths of the soil profiles are smaller than tillage depths applied by farmers, parent materials or mudstone bedrocks would be crushed by tillage operations to retain a certain cultivated depth [5,6,9,12]. The introduction of bedrock fragments into the soil not only promotes its weathering into soil within a relatively short period but also leads to a shorter soil renewal cycle, resulting in significant differences in soil properties compared to the original soil [13].

Additionally, under the transport effects of cultivation, bedrock fragments gradually displace horizontally and vertically, leading to changes in the thickness of the soil–rock mixture, fragment content, and vertical gradient of bedrock content. This, in turn, significantly affects the soil structure, physicochemical properties, and hydrological processes on slopes [13,14,15,16,17]. In the region of purple soil from Sichuan, where rainfall is abundant, and slopes are cultivated, soil erosion on sloping farmlands is not only influenced by bedrock fragments but also regulated by rainfall [18]. Furthermore, variations in rainfall intensity during natural rainfall events are common. However, research on the impact of rainfall intensity variations (rainfall patterns) on the soil transport process on slopes is scarce. Therefore, it is essential to study the effects of bedrock fragments on soil erosion characteristics in purple soil croplands under different rainfall patterns to address these knowledge gaps.

This study aims to investigate the influence of the cultivation-induced bedrock fragments under different rainfall patterns on soil erosion and hydrological characteristics in purple soil croplands. To achieve this, artificial rainfall experiments were conducted with the following three objectives: First, to understand how bedrock fragment affects the hydrodynamic parameters on the slope surface under different rainfall patterns. Second, to examine the variations in runoff and sediment yield from slopes with bedrock (WB) fragments under different rainfall patterns and rainfall sequences during the soil erosion process. Third, to explore differences in the critical dynamic conditions for soil erosion and soil erodibility between WB and without bedrock (CK) on slopes. This research has significant implications for the development of high-intensity soil conservation measures for downhill lands, controlling soil loss on sloping farmlands, and promoting agricultural economic development in the region.

2. Materials and Methods

2.1. Experimental Site

To investigate the impact of incorporating crushed bedrock into the soil on slope hydrological processes, the soil samples used in the experiment were collected from Qianjin Township, Ya’an City, Sichuan Province (30°3′58″ N, 103°11′52″ E). This region is located in the western part of the Sichuan Basin and experiences a subtropical monsoon-influenced humid climate. It is renowned for its high annual rainfall, with an average annual precipitation of 1455.3 mm. Precipitation is concentrated mainly in the summer, followed by the spring and autumn seasons. The average annual temperature is 15.9 °C, with wind speeds ranging from 0.8 to 1.2 m s⁻¹ and an average relative humidity of 82%. Ya’an city is characterized by numerous mountain ranges and rivers, with elevations ranging from 627 to 5793 m, resulting in significant topographical variations. The cropping system in this area follows a biannual cultivation cycle, primarily focused on the cultivation of crops and economic plants such as loquat (Eriobotrya japonica (Thunb.) Lindl.), walnut (Juglans regia L.), camellia (Camellia japonica L.), and gastrodia (Gastrodia elata Bl.).

The soils in this area, formed from purple mudstone/shale of Jurassic age, are classified as Eutric Regosol according to the FAO soil classification system. The soil texture is predominately loam (USDA classification) with 42.62% clay, 28.33% silt, and 29.05% sand. The soil profiles are relatively shallow, with thin topsoil layers directly overlying the bedrock. Soil samples were collected from a depth of 0 to 30 cm. After removing impurities such as stones and roots from the soil samples, they were air-dried and sieved through a 10 mm sieve. Due to the ease of mechanical fragmentation of purple bedrock (especially under alternating wet and dry conditions) and the difficulty in accurately controlling the content, particle size, and distribution of bedrock fragment during the experiment, this study chose to use hard gravel (density: 2.74 ± 0.05 g cm⁻³) as a substitute for purple parent rock debris (density: 2.10 ± 0.05 g cm⁻³) in the simulation experiments.

2.2. Experimental Design

A movable hydraulic lift test trough with slopes ranging from 0 to 30 degrees was used. The steel tanks had dimensions of 4.5 m in length, 2 m in width, and 0.5 m in depth. The rainfall simulator used in the experiments was the NYJL-10 artificial rainfall system (Nanjing Nanlin Electronic Technology Co., Ltd., Nanjing, China, Figure 1). It consisted of 15 sets of nozzles, with each set containing 7 nozzles. Rainfall was simulated from below, with a rainfall height of 6 m, and rainfall intensity ranging from 10 to 240 mm h⁻¹. Through repeated experiments, the effective rainfall area of the system was determined to be 10 × 6 m², with a rainfall uniformity of ≥88%.

The steel tanks were lined with mesh fabric that had circular holes (diameter: 10 mm) at a rate of 100 holes per square meter, simulating a weakly permeable layer on natural slopes. Four sizes of hard gravel (5–10, 10–20, 20–40, and 40–80 mm) were selected and mixed in a 1:1:1:1 ratio. The mass of soil and gravel required for each treatment was calculated based on the bulk density and moisture content of air-dried soil, as well as the density of the gravel (with a gravel mass content of 20% and a volume content of 12%). The gravel and test soil were thoroughly mixed in a drum-type mixer (220 V electric) and then layered into the steel tanks, compacted to a thickness of 40 cm (WB). A control group was included using the test soil without added gravel (CK). During the compacting process, the gravel and soil layers were packed in 5 cm increments, and each packed soil layer was raked lightly before the next layer was packed to ensure uniformity and continuity in the soil structure. The soil amount of each layer was kept as constant as possible to sustain similar bulk density and uniform spatial distribution of natural soil particles.

To reduce the surface variability of the experimental soil and ensure consistent initial soil moisture content before each test, preliminary rainfall was conducted at a rainfall intensity of 15 mm h⁻¹ until runoff was about to occur on the slope. Low-lying areas were filled based on water accumulation conditions to reduce slope undulations. The soil reaching a saturated state was covered with plastic sheeting, and the trough was raised to a 15-degree slope. After a 12 h rest period, artificial rainfall experiments were conducted. Following each formal rainfall event, the slope surface was raked, air-dried soil was replenished, and the process was repeated.

Based on the analysis of natural rainfall processes in typical purple soil areas, five rainfall patterns were designed for the study: decreasing pattern (rainfall intensity sequence as 150–120–90–60–30 mm h⁻¹, Dec), increasing pattern (rainfall intensity sequence as 30–60–90–120–150 mm h⁻¹, Inc), Constant pattern (rainfall intensity distribution with 90 mm h⁻¹, Con), increasing–decreasing pattern (rainfall intensity sequence as 30–90–150–90–30 mm h⁻¹, Inc-dec), and decreasing–increasing pattern (rainfall intensity sequence as 150-90–30–90–150 mm h⁻¹, Dec-inc). Each rainfall pattern had a duration of 12 min, totaling 60 min for Dec, Inc, and Con rainfalls. However, for Inc-dec and Dec-inc patterns, the durations were 13.8 min and 10.6 min, respectively, totaling 69 min and 53 min. This ensured that the total rainfall for each rainfall pattern was 90 mm h⁻¹ (as detailed in Figure 2). Runoff initiation times were recorded when noticeable surface flow occurred on the slope, and stable water flow was observed at the outlet. Samples of surface runoff were collected in a 3 min interval into metering plastic buckets to measure the volume and mass of the runoff–sediment mixture. The collection ended when no runoff occurred. Each sampling duration was 20 s. Conventional methods were used to measure hydraulic parameters, such as flow velocity (v, m s⁻¹), surface width (w, m), depth (d, m), and water temperature (t, °C). v was determined at the upper, middle, and lower slope positions using means of KMnO₄ solution during the simulated rainfall event, after the flow discharge stabilized. The upper, middle, and bottom slope positions were marked from 0 m to 3 m, from 3 m to 6 m, and from 6 m to 8 m within the runoff plots, respectively. Each treatment was conducted two times. The collected sediment samples were dried for calculating runoff rate and sediment yield.

2.3. Hydraulic Parameters

To analyze the hydraulic characteristics of the flow, calculations were performed to determine the Reynolds number (Re), Froude number (Fr), Darcy–Weisbach resistance coefficient (f), effective shear stress (τ, Pa), unit stream power (p, m s⁻¹), and soil detachment rate (Dr, g m⁻² S⁻¹). The formulas for calculating these parameters can be adopted in this study and derived as follows [19,20,21,22,23]:

Re = vR/η

(1)

$$Fr=v/\sqrt{gh}$$

(2)

F = 8gRJ/v²

(3)

τ = γgJ

(4)

p = vJ

(5)

Dr = Kr(τ–τ_c)ⁿ

(6)

where R is the hydraulic radius (m), r = A/Wp; A is the cross-sectional area, A = w × d; Wp is the wetted perimeter (m), Wp = w + 2d; η is the water kinematic viscosity (m² s⁻¹), η = 0.01775/(1 + 0.0337t + 0.000221t²); g is the gravitational acceleration (9.8 m s⁻²); h is the depth of runoff (m), and h was calculated with the equation employed by Wang et al. [23]; γ is the volumetric weight of water (N m⁻³); J is the hydraulic gradient (m m⁻¹), which is approximately substituted by the sine of slope gradient; Kr is the erodibility factor (g N⁻¹ min⁻¹); τ_c is critical shear stress (Pa); n is an exponent. Kr, τ_c, and n can be obtained through regression analysis between Dr and τ. More detailed procedures of above parameters can be found in the literature by Su et al. [20] and Wang et al. [23].

2.4. Statistical Analysis

Paired-sample t-tests were used to analyze the significance of differences in Darcy–Weisbach friction factor between WB and CK under different rainfall patterns with IBM SPSS Statistics 26.0 software (SPSS. Inc., Chicago, IL, USA). The differences were assumed to reach a statistically significant level at p < 0.05. Origin 2021b (Origin Lab. Inc., Northampton, MA, USA) software was used to fit the equations of Dr and τ.

3. Results

3.1. Variations in Characteristics of Hydrodynamic Parameter

3.1.1. Flow Velocity

From Figure 3, it can be observed that the variation patterns of v values on the WB and CK are similar at different slope positions. Under the Dec pattern, the lowest average v value on the WB occurs at the middle slope position (0.08 m s⁻¹), whereas the peak value occurs at the lower slope position (0.10 m s⁻¹). For the other rainfall patterns, both WB and CK exhibit an increasing trend in average v values from the upper to the bottom slope position, with increases ranging from 21.21% to 53.39% for the WB and from 28.79% to 51.60% for the CK. This indicates that slope position has a significant impact on runoff velocity under various rainfall patterns.

Furthermore, different rainfall patterns have a considerable influence on the v values for both WB and CK. The highest and lowest values occur under the Con and Dec patterns, respectively, with the average v value under the Con being 28.49% higher (WB) and 20.76% higher (CK) compared to the Dec. On different soil types, the variation range of average v values for the CK is between 0.06 and 0.08 m s⁻¹, whereas the v values on the WB increase by 23.34% to 48.94% compared to that of the CK. This result indicates that bedrock fragments have a promoting effect on runoff velocity during rainfall.

3.1.2. Reynolds Number and Froude Number

The changes in runoff patterns are plotted in Figure 4, showing significant differences in runoff patterns for both WB and CK between different slope positions. Moreover, as runoff converges, the overall runoff patterns between the upper and lower slope positions transition from “laminar flow-slow flow zone” (Re ≤ 500, Fr ≤ 1) to “laminar flow-rapid flow zone” (Re ≤ 500, Fr > 1). Specifically, at the upper slope position, except for the WB under the Con pattern, which exhibits runoff patterns in the “laminar flow-rapid flow zone”, all other slope surfaces are in the “laminar flow-slow flow zone”. In the middle slope position, under the Dec-inc pattern, the runoff patterns for the WB gradually transition into the “laminar flow-rapid flow zone” on slopes. When runoff converges at the lower slope position, except for the CK under the Dec, which remain in the “laminar flow-slow flow zone”, the runoff patterns on other slopes transition from the “laminar flow-slow flow zone” to the “laminar flow-rapid flow zone”. This indicates that under the conditions of this experiment, rainfall patterns and bedrock fragments jointly control the runoff patterns on slopes.

3.1.3. Darcy–Weisbach Resistance Coefficient

As shown in Figure 5, the f value generally decreased gradually from the upper slope to the bottom slope, with only the middle slope position of the WB under the Dec pattern having a higher f value compared to the upper and bottom slope positions. Except of Inc-dec pattern, there were significant differences in the f values between WB and CK for varying-intensity rainfall patterns (P < 0.05). Numerically, the average f value at the bottom slope position is reduced by ranging from 42.38% to 72.80% compared to the middle and upper slope positions. When comparing different rainfall patterns, the f values under the Dec pattern are significantly higher than under the other rainfall patterns (WB: 19.25% to 135.83%; CK: 7.22% to 79.85%). The finding indicates that surface runoff generated by the Dec has the weakest erosion capability. Both WB and CK, bedrock fragments can increase surface roughness to reduce runoff erosion, with the average resistance coefficient being increased by 1.59% to 53.16% (mean 33.45%), compared to that of the CK. This suggests that both rainfall patterns and bedrock fragments significantly affect soil’s resistance coefficient.

3.1.4. Flow Shear Stress

Figure 6 reflects the changes in τ for both WB and CK under different rainfall patterns. It can be observed that the variation in runoff shear force among different slope positions is similar to that of the resistance coefficient, showing an overall trend in decreasing τ values from the upper to the bottom slope positions, with reductions ranging from 16.80% to 34.99% (WB) and 22.42% to 34.02% (CK). Furthermore, among different rainfall patterns, the WB have the maximum and minimum τ values of 0.18 Pa (Dec) and 0.14 Pa (Con), respectively, whereas τ values under the other rainfall patterns are similar, ranging from 0.16 Pa to 0.17 Pa. For the CK, the τ values in descending order are Dec-inc (0.22 Pa) > Dec (0.21 Pa) > Inc (0.19 Pa) > Con (0.18 Pa) > Inc-dec (0.16 Pa). The influence of bedrock fragments on runoff shear force on the soil should not be underestimated. Under different rainfall patterns, the WB has significantly lower τ values compared to the CK (reduction ranging from 1.52% to 22.19%, mean 13.43%). This result indicates that bedrock fragments inhibit the erosion of surface runoff on the soil, and bedrock fragments are the primary factor affecting flow shear force.

3.1.5. Unit Stream Power

The overall trend in changes in the p is that the further away from the top position, the larger the p value (increase ranging from 20.21% to 53.39%, Figure 7). This trend is similar to the variation in v values. Moreover, under different rainfall patterns, the changes in p values on the WB and CK are relatively gradual, ranging from 0.020 m s⁻¹ to 0.025 m s⁻¹ and 0.017 m s⁻¹ to 0.021 m s⁻¹, respectively. However, the p values on the WB are still 3.86% to 27.86% higher than those on the CK. This shows that bedrock fragments have a better energy dissipation effect and can effectively resist the transport and soil erosion caused by runoff. Additionally, regardless of WB and CK, the maximum and minimum p values occur under Dec and Con patterns, respectively, the p value under Dec being 1.21 to 1.28 times than that of under Con. This demonstrates that rainfall patterns have a significant impact on unit stream power.

3.2. Runoff Volume and Sediment Yield

Surface runoff on slopes is strongly influenced by the type of rainfall (Figure 8). Total runoff volume ranked similarly as Dec > Dec-inc > Con > Inc > Inc-dec and Dec-inc > Con > Dec > Inc > Inc-dec for the CK and WB, respectively. The sediment yield was ranked in the following order for various rainfall patterns Dec > Inc > Dec-inc > Inc-dec > Con for the WB, yet it followed the pattern Con > Dec > Inc > Inc-dec > Dec-inc for the CK. Furthermore, it was observed that only under the Con pattern, bedrock fragments had a significant inhibitory effect on soil erosion and sediment yield, reducing sediment yield by 45.41%, compared to the CK. However, under the other rainfall patterns, the WB showed a significant increase in both runoff volume and sediment yield, with increases by an average of 59.29% for runoff volume and 71.62% for sediment yield. This indicates that under variable rainfall patterns, bedrock fragments can markedly enhance both surface runoff and sediment yield on slopes.

Due to the limited occurrence of rainfall intensity at 60 and 120 mm h⁻¹ in this study, with only one instance each in the Dec and Inc patterns, the sample capacity is insufficient and not representative. Therefore, this study focuses on discussing the impact of rainfall sequence on runoff volume and sediment yield for rainfall intensities of 30 mm h⁻¹, 90 mm h⁻¹, and 150 mm h⁻¹. To quantify the net contributions of bedrock on runoff volume and sediment yield for each rainfall intensity and stage, a contribution rate, defined as the ratio of (Runoff volume/Sediment yield _WB − Runoff volume/Sediment yield _CK) to the Runoff volume/Sediment yield _WB, was introduced in this study.

It is evident that rainfall sequence at the same rainfall intensity has a significant influence on surface runoff and sediment yield for both soil types under various rainfall patterns (

Table 1. Comparison of runoff volume and sediment yield under different rainfall patterns and stages.

Rainfall Intensity (mm h−1)	Rainfall Pattern	Stage	Runoff Volume (L)			Sediment Yield (g m−2)
			WB	CK	Contribution Rate (%) 1	WB	CK	Contribution Rate (%)
30	Dec	V	33.21	11.44	11.88	19.36	3.07	5.36
	Inc	I	8.56	5.96	1.63	14.41	0.99	4.56
	Inc-dec	I	15.20	6.39	6.23	30.85	13.11	6.79
	Inc-dec	V	17.33	12.75	3.24	18.48	7.21	4.31
	Dec-inc	III	50.73	31.06	8.87	12.38	8.55	2.02
90	Dec	III	77.78	23.71	29.50	84.69	44.90	13.08
	Inc	III	57.86	24.10	21.18	106.41	39.90	22.60
	Inc-dec	II	73.05	41.16	22.54	222.56	111.33	42.58
	Inc-dec	IV	49.63	29.66	14.11	48.23	23.93	9.30
	Dec-inc	II	64.76	47.76	7.67	88.19	31.77	29.71
	Dec-inc	IV	34.72	26.48	3.72	37.25	16.43	10.96
	Con	I	45.65	31.43	6.96	89.28	189.72	−21.48
	Con	II	58.89	41.87	8.34	50.29	111.36	−13.06
	Con	III	58.99	43.63	7.52	38.62	60.72	−4.72
	Con	IV	59.28	42.76	8.09	37.18	52.09	−3.19
	Con	V	61.23	44.50	8.19	39.91	53.79	−2.97
150	Dec	I	79.23	65.87	7.29	279.99	159.71	39.55
	Inc	V	94.03	66.43	17.31	204.54	160.47	14.97
	Inc-dec	III	93.21	51.52	29.47	146.44	105.64	15.62
	Dec-inc	I	64.21	39.46	11.17	195.13	61.70	70.26
	Dec-inc	V	96.14	76.89	8.69	156.02	71.47	44.52

¹ Contribution rate is the ratio of (Runoff volume/Sediment yield _WB—Runoff volume/Sediment yield _CK) to Runoff volume/Sediment yield _WB.

). For instance, at a rainfall intensity of 30 mm h⁻¹, the maximum and minimum runoff volume occur at stage III in the Dec-inc and stage I in the Inc, respectively, with differences ranging from 4.21 to 4.93 times. The maximum sediment yield occurs at stage I in the Inc-dec (WB: 30.85 g m⁻², CK: 13.11 g m⁻²). At a rainfall intensity of 90 mm h⁻¹, the maximum runoff volume in the WB and the minimum runoff volume in the CK both occur at stage III in the Dec, with slopes containing bedrock having 2.28 times higher runoff volume. Sediment yield, under variable rainfall patterns, reaches its maximum at stage II in the Inc-dec, with increases ranging from 43.32% to 497.48%. At 150 mm h⁻¹ rainfall intensity, the largest difference in runoff volume was observed at stage V and stage I in the Dec-inc, with the former being 49.73% to 94.86% higher.

Furthermore, it is noted that different rainfall sequences exhibit significant differences in runoff volume and sediment yield under the same rainfall patterns. For instance, the Inc-dec, Dec-inc, and Con patterns under a rainfall intensity of 90 mm h⁻¹ as well as the Dec-inc pattern under 150 mm h⁻¹ rainfall, show an overall decrease in runoff volume and sediment yield with increasing rainfall stages. This indicates that runoff generation and sediment production exhibits distinct temporal variability under different rainfall patterns. Additionally, except for the Con pattern, for the same rainfall intensity and different rainfall sequences, the WB consistently yield higher runoff volume and sediment yield compared to the CK, with a contribution rate ranging from 1.63% to 29.50% and 2.02% to 44.52%, respectively. The results indicate an increasing trend in average contribution rate among the three intensities: 6.37% and 4.61% for 30 mm h⁻¹, 12.53% and 7.53% for 90 mm h⁻¹, as well as 14.79% and 36.98% for 150 mm h⁻¹, respectively, for runoff volume and sediment yield. This implies that the net effect of bedrock impact increased with increase in rainfall intensity.

3.3. Soil Erodibility and Critical Shear Stress

Kr and τ_c were calculated from Equation (6) by a power function equation and were compared with the observation. The results, as shown in Figure 9, reveal that the dynamic critical conditions for soil detachment rate on the WB and CK were 0.49 Pa and 0.77 Pa, respectively. The WB consistently exhibited lower τ_c values compared to the CK. Additionally, the soil erodibility parameters for the WB (Kr = 0.23 g N⁻¹ min⁻¹) were greater than those for the CK (Kr = 0.07 g N⁻¹ min⁻¹). It is important to note that the model fitted was quite high in both soil types (P < 0.05), with R² values exceeding 0.80, indicating that bedrock fragments exacerbate soil detachment regardless of soil types.

4. Discussion

The runoff volume and sediment yield on the WB and CK exhibited distinct temporal patterns under varying rainfall intensities. Among the variable rainfall patterns, the WB consistently exhibited their maximum runoff volume and sediment yield under Dec and Inc-dec patterns, whereas the minimum values occurred under Dec-inc pattern. This isattributed to the fact that the Inc-dec pattern had a lower initial rainfall intensity of 30 mm h⁻¹, significantly lower than the initial rainfall intensities of other rainfall patterns. As a result, the soil absorbed a substantial amount of runoff during rainfall, leading to delayed soil saturation and surface runoff generation. Moreover, the influence of rainfall patterns on soil erosion may involve various processes such as changes in microtopography and runoff generation mechanisms [24,25,26,27,28]. Under Inc rainfall conditions, the lower initial rainfall intensity resulted in lower kinetic energy of raindrops, which led to reduced detachment and transport capacity for sediment. Consequently, only loosely packed surface areas were eroded, resulting in lower runoff volume and sediment yield [29]. In contrast, the Dec pattern had higher initial kinetic energy, which had a more pronounced effect on altering surface microtopography. In the later stages of rainfall, when the soil on the slope became saturated, the transport of sediment was primarily influenced by the detachment and transport capacity of runoff, leading to variations in runoff volume and sediment yield with changing rainfall intensity [30]. For soils like purple soil, which have a dual-layer structure with a shallow upper soil layer and underlying weathered parent rock and are susceptible to internal erosion and deep infiltration, the impact of rainfall patterns on erosion becomes more apparent. This is because soil texture and soil structure play a significant role in the erosion process, and the influence of rainfall patterns is more pronounced in such soils [31].

This study has found that rainfall patterns can significantly influence a series of hydraulic parameters. In the case of variable rainfall patterns, for different soil properties on slopes, the f and p values are highest under the Dec pattern, whereas v and τ values are relatively smaller. This phenomenon can be attributed to the significant impact of the Dec pattern on the surface microtopography in the early stages. The surface soil becomes relatively smooth under this pattern due to its high initial kinetic energy and pronounced splashing effect on the soil surface, resulting in a high surface roughness with the prolonged rain event [32]. The high rainfall intensity creates the immediate ponding, and this prevents the soil air from escaping and thus slows water infiltration. Additionally, the runoff generation in the intensified rainfall pattern occurs later and at lower flow rates. The erosion process on the soil surface is slower under this pattern. Consequently, the runoff remains in a “laminar-slow flow” state, resulting in weaker soil erosion.

It was observed that the presence of bedrock fragments by long-term tillage has a significant promoting effect on surface runoff and sediment yield on slopes. This is partly because bedrock fragments in the soil primarily contribute to enhancing the overall soil erosion resistance, thereby reducing soil loss and lowering erosion rates [10,33], and partly because bedrock fragments, in this case, lead to significant changes in soil physical properties and affect soil structure and infiltration rates [34,35]. However, whether this impact increases or decreases runoff volume and sediment yield depends largely on the distribution of bedrock fragments within the soil. Rock fractures with multiscale behaviors and high anisotropy natures have been regarded as preferential channels for runoff during rain events, making a great difference in the hydraulic conductivity of bedrock fragments [36]. As the bedrock distribution depth increases, they can promote the formation of preferential flow pathways, improving water infiltration but also increasing soil instability on slopes [37]. Additionally, bedrock distributed deep within the soil can lengthen the movement pathway of soil moisture, obstruct surface water flow, raise groundwater levels, increase pore water pressures, and exacerbate surface water and soil loss [38]. This leads to thicker water layers, faster flow velocities, and increased erosive potentials of runoff on slopes, resulting in higher sediment yields [13,39].

WB exhibited significantly higher flow velocities, resistance coefficients, and unit stream powers while having notably reduced runoff shear forces compared to the CK. It is evident that bedrock fragments have a substantial impact on the hydrological characteristics and erosion processes. In general, bedrock fragments distributed within the soil mainly consist of four types: covering the topsoil, partially embedded within the topsoil, completely mixed within the topsoil, and incorporated deep within the soil profile. These different positions have varying effects on hydrological characteristics, such as affecting overland flows, water movement pathways, pore characteristics (e.g., porosity, effective size, pore distribution, and connectivity), and altering hydraulic properties [40,41,42]. The uniform distribution of bedrock fragments in this study might lead to the potential mutual compensation of their impacts on hydrological characteristics, adding to the uncertainty in runoff and sediment transport processes.

Bedrock fragments within the soil surface may increase surface roughness, hindering surface water flow. Simultaneously, they may change the distribution of larger soil pores and pore connectivity, promoting subsurface flow while reducing surface runoff generation. This dynamic contributes to the observed increase in resistance coefficients and reduction in runoff shear forces [13,30,43]. In summary, the presence of bedrock within the soil significantly impacts the hydrological characteristics and erosion processes. However, the influence of bedrock fragments on hydrological characteristics is very complex, and it is due to the combined influence of soil properties, configuration, size, position and rainfall intensity, which is difficult to describe quantitatively. Further research needs to quantitatively or qualitatively consider the influence of environment, soil properties, bedrock properties, rainfall, and other factors.

5. Conclusions

The impacts of bedrock fragment on surface runoff and soil erosion were quantitatively studied using simulating five rainfall patterns (Dec, Inc, Inc-dec, Dec-inc, and Con), each event with a different temporal rainfall pattern but all delivering the same total precipitation. Significant differences were observed in runoff and sediment between WB (soil–bedrock mixtures) and CK (pure soil), and soil losses from both WB and CK were significantly affected by rainfall patterns and stages. Under the same total precipitation, the WB shows both maximum runoff volume and sediment yield under the decreasing rainfall pattern, whereas the CK exhibits maximum runoff and sediment yield under the Dec-inc and Con patterns, respectively. The sediment yield was ranked in the following order for various rainfall patterns Dec > Inc > Dec-inc > Inc-dec > Con for the WB, yet follows the pattern: Con > Dec > Inc > Inc-dec > Dec-inc for the CK. As runoff converged, significant changes in the flow regime were found between the upper and bottom slope positions for both WB and CK. The v and p values under different rainfall patterns presented an increasing trend from the upper to the bottom slope positions, yet the f and τ values displayed the opposite trend for both WB and CK. A fitted equation also showed that the WB had lower critical kinetic conditions and higher soil erodibility, exacerbating runoff generation and sediment yield. Additionally, under different rainfall patterns, the WB generally exhibited higher flow velocities, resistance coefficients, and unit stream power compared to the CK. In general, the runoff volume and sediment yield for the WB and CK exhibited pronounced temporal variations under different rainfall intensities. It is evident that rainfall pattern plays a crucial role in the hydrological characteristics and erosion processes between the WB and CK. There is still a substantial knowledge gap between fundamental erosion mechanisms and process-based erosion prediction models. We hope that the results presented here are helpful in building a better understanding of the effects of rainfall patterns on the bedrock erosion process, but further research is yet merited.