Transmission Characteristics of the Macropore Flow in Vegetated Slope Soils and Its Implication for Slope Stability

Abstract

Macropores in the soil of vegetated slopes greatly affect the rainfall infiltration process. In this paper, a realistic 3D macropore network model of a soil column sample is established by CT scanning. Then, the transmission process of the macropore flow is simulated based on MODFLOW. The results show that (1) the shapes of macropores in the soil contain not only the dominant proportion of the circular tube but also a small proportion of the flake. (2) The velocity of macropore flow has a maximum of up to 0.2~0.3 m/s, which is much higher than that of matrix flow. In every single macropore, the flow velocity is the greatest at the central axis perpendicular to the extension and at the throat along the extension. (3) Due to the development of the macropore network system, rainwater can quickly pass through the soil profile in the form of preferential flow or pipe flow, which shortens the lag time of the peak discharge response to rainfall. This process can free up space for the next recharge, but reduce the overall quality of the soil, laying the foundation for the slope failure. Our work helps to unravel the mechanism of rainfall-induced landslides and promote harmony and sustainable development between humans and nature.

1. Introduction

Rainfall-induced landslides constitute one of the most widely distributed types of slope geological hazards in the world (e.g., refs. [1,2,3,4,5,6,7,8,9,10]), which, essentially, is a response behavior of slope stability to groundwater [11]. Rainfall may pose a threat to slope stability only when it infiltrates and transforms into groundwater, causing a groundwater table of sufficient proportional thickness [12]. This process is so-called rainfall infiltration [13]. Clarifying this process in vegetated slope soils is not only crucial for revealing the mechanism of rainfall-induced landslides but also helps to enhance our ability to defend against landslide disasters, promoting harmony and sustainable development between humans and nature.

The intensity of rainfall infiltration depends on the soil structure and hydraulic and physical properties, such as specific moisture capacity, moisture content, permeability, etc. The main factor determining these properties is the soil porosity, especially the macro-porosity.

Macropores, with the equivalent diameter of >1 mm [14,15,16], are a common structure in vegetated slope soils. Generally, the types of macropores can be classified as biological macropores [17], which mainly include root–soil interstices, rotten root channels, and worm pores, and structural macropores, which mainly contain inter-aggregate pores, soil–gravel interstices, and soil cracks [11]. Regardless of the type, macropores allow water to flow freely under gravity without being affected by capillary forces [18] and, as such, play an important role in the rainfall infiltration process. For example, Mosley [19] analyzed a large amount of rainfall data and concluded that macropore flow is the main mechanism for the formation of channel stormflow, and water could move through macropores (mainly root channels) at a rate twice the infiltration rate of the soil. Beven and Germann [20] deemed that a small proportion of macropores in soil porosity determined the flow velocity in soils. Cey and Rudolph [21] studied macropore flow processes in saturated soils under in situ conditions, noting that macropores are distinguished from the soil matrix in that they allow water and contaminants to pass through the soil profile in the form of preferential flow and that they occupy a very important position in the hydrological system. Qin et al. [22] denoted that the macropore network system formed by rotten root channels might include entire hillslopes and provide a rapid drainage path for water following rainfall.

In recent years, numerical simulations have become a convenient way to study macropore flow due to the rapid development of computing hardware and software technology. More importantly, it also offers the possibility of quantitatively understanding the macropore flow. Christiansen et al. [23] simulated the macropore flow and transport processes at the catchment scale and noted that macropores could rapidly transport most of the infiltrating water and solutes from the plough pan at a 20 cm depth some distance downwards before it flowed back into the soil matrix. Köhne and Mohanty [24] used HYDRUS2D software to simulate the circular-macropore flow in a soil column and found that the velocity of macropore flow was hundreds of times higher than that of matrix flow. Alaoui [25] simulated the macropore flow using the dual-permeability MACRO model and manifested that the volume of water that flowed from macropores and was expected to reach groundwater varied between 81% and 94% in brown soils. Zhang et al. [26] pointed out that soil macropores, as preferential flow pathways, could rapidly transport water, air, and chemicals through the soil when they simulated the effect of macropore morphology on water infiltration by using COMSOL.

To date, the macropore models in previous studies are either ideal, of which the macropores are simplified as circular tubes, or randomly generated and could not actually incorporate the macropore distribution characteristics and patterns. The use of 3D numerical simulation to reveal the transmission characteristics of the macropore flow is seldom studied due to the difficulty of constructing the real macropore network system and generating meshes.

In this paper, the internal structure of a soil column sample in the study area will be obtained by means of CT scanning, so that a realistic 3D macropore network model can be established; through the MODFLOW seepage calculation program, the macropore flow simulation will be carried out in order to reveal the transmission characteristics of the macropore flow in the soil of vegetated slopes; finally, the implication of the rapid transmission of macropore flow for the slope stability is discussed.

2. Materials and Methods

2.1. Study Area

The Touzhai Valley watershed (Figure 1, 27°32′52″~27°34′15″ N, 103°51′09″~103°52′50″ E) is located in the northeast of Yunnan province, China, about 320 km from Kunming. The climate is defined as a continental monsoon plateau, with a subtropical and warm temperate zone co-existing in the northern latitude. Elevations of the watershed range from 1820 to 2940 m a.s.l., with a 3.9 km long channel and a drainage area of 3.2 km² (Figure 1). According to the data from Wang et al. [27], the average annual precipitation was 1082.8 mm, mainly concentrated from May to September, with a proportion of 75%; the highest and lowest temperatures in the upper reaches were 27.1 °C and −19.3 °C, respectively, with the average of 8.2 °C. More than 90% of the area is covered by vegetation. Through field investigation, 28 vegetation species were identified, which mainly belong to Rosaceae, Asteraceae, Berberidaceae, Salicaceae, +Hamamelidaceae, etc.

At about 6:00 p.m. on 23 September 1991, the Permian basalts in the upper reaches of the Touzhai Valley, located at an altitude of 2300~2580 m a.s.l., failed. The mass of about 9 × 10⁶ m³ was ejected from the source area, hit the opposite valley wall, rushed down the valley, and finally stalled at the left bank of the Pan River. The traveling distance was about 3.4 km and lasted for about 3 min, covering all of the Touzhai Village located at the mouth of the valley, causing 216 deaths and more than RMB 12 million of direct economic losses.

2.2. Soil Sampling

The soil column sample used for numerical model construction is derived from the left bank of the landslide initiation zone, with underlay by Permian Emeishan basalt (P₂β) (Figure 1). The sampling process is as follows:

(1) An area of 1 × 1 m (arbors should be kept separate so that the sampling process would not be affected by their coarse roots) was delineated, and subsequently, the litter layer, humus layer, and other impurities on the surface were cleaned up, while a 30 × 30 cm sampling area was designated (Figure 2a); (2) the soil around the sampling area was excavated with a height of 75 cm soil column left, and then all the surfaces were trimmed to make it as flat as possible (Figure 2b); (3) the soil column was wrapped with multiple layers of preservative film inside and cotton cloth outside to reduce moisture loss and vibration during the transit (Figure 2c); (4) finally, the rough size of 28 × 28 × 50 cm for the soil column sample was fixed with customized wooden planks and prepared for scanning (Figure 2d).

2.3. Computed Tomography Scanning and Processing

Computed tomography (CT) is a technique that utilizes precisely collimated X-ray beams, gamma rays, ultrasonic waves, etc., together with highly sensitive detectors to perform sectional scans. Due to its characteristics of non-invasiveness, short duration, and clear images, it has been widely applied in the study of macropores [28,29,30,31,32].

The CT scanning equipment used here was Somatom Sensation Open 40 (Figure 3a). The scanning voltage, current, thickness, and size were 120 kV, 176 mA, 1.5 mm, and 484 × 484 mm, respectively (Figure 3b). The slice accuracy was 512 × 512 pixels with a single size of 0.945 × 0.945 × 1.5 mm. A total of 334 sections were obtained (a depth range of 0 to 500 mm, and the top surface of the soil column was specified as Z = 0). However, 267 complete and effective sections (a depth range of 100 to 500 mm) were finally selected due to the inevitable damage to the corners of the sample during transportation. All of these slice images (in DICOM format and a color depth of 16-bit) were imported into the software of Materialise’s Inter-active Medical Image Control System (MIMICS version 17.0) developed by Materialise Company for subsequent processing.

Different materials in the soil were separated according to their linear attenuation coefficient to X-ray (i.e., CT value, Hounsfield Unit, HU for short). In MIMICS, it is not necessary to pay attention to the specific CT values of materials. Only a section was selected and marked with the tool of the profile line for macropores, the matrix, and gravel (Figure 4a,b), and the software was able to automatically read their CT values. This operation was repeated several times, and the CT value range of a certain material could be seen from the CT value distribution diagram. As can be seen from Figure 4c, the CT value ranges of macropores, matrix, and gravels are −1024~−424 HU, 26~476 HU, and 1376~2126 HU, respectively. In addition, due to the fact that the edges of the soil sample were not neat and uniform, the slices needed to be cut. Finally, the available size of every slice was determined to be 25 × 25 cm (Figure 4a,b).

2.4. MODFLOW Simulation Program

MODFLOW is a modular open-source computing program developed by the U.S. Geological Survey [33]. It has been specifically used in the modeling and prediction of groundwater conditions, groundwater/surface-water interactions, moisture and solute transport in porous media, variable-density flow, aquifer-system compaction, land subsidence, parameter estimation, and groundwater management.

MODFLOW is composed of a main program and a series of attached subroutines. The subroutines are independent and interoperable. During operation, the whole process of simulation can be divided into multiple stress periods, and each stress period can be divided into multiple time periods. The head at the end of each time period can be obtained by an iterative solution of the finite difference equation (FDE).

3. Model Construction

3.1. Numerical Model

Figure 5 showed the separation and recombination of macropores, matrix, and gravel in the soil column sample from different angles. The distribution of macropores in the soil was intricate (Figure 5a). Generally, most macropores were the shape of circular tubes, which should be biological macropores. They were not coordinated except for in the direction of horizontal or extending obliquely to the deep. The maximum equivalent diameter of the pores was 4 cm. Below the depth of Z = 25 cm (Figure 5(a-2)), the proportion of circular tubes was significantly reduced, and the shape began to change into flakes, which should be structural macropores. In order to more quantitatively understand the distribution trend of macropores in the sample, the area ratios of macropores (namely macro-porosity) in some sections were calculated (Figure 6). It is clear that, on the whole, macro-porosity decreased with the increase in depth. The macro-porosity at the top of the soil column (Z = 0) was 28.08%, which then greatly dropped to 11.37% at the bottom (Z = 399.0 mm).

The gravel was scattered in the matrix without obvious features (Figure 5b,d). However, it can be roughly observed that the size of the gravel tended to increase with the increase in depth, and a boulder appeared at the bottom. Similarly, in order to quantify the variation of gravel sizes with depth, the long axis lengths of gravel were used to represent their sizes, and the Z coordinates of their centers were used to represent their depths (see Figure 6). Within the range of <200 mm, gravel sizes were mostly concentrated between 3.5 and 350 mm, while within the range of >200 mm, gravel sizes augmented and began to decentralize. The long axis length of the boulder mentioned above at the bottom reached about 142.3 mm.

The distribution pattern of the matrix was not apparent (Figure 5c). However, according to the distributions of macropores and gravel, the matrix should decrease with increases in depth. Moreover, the slope soil thickness of the Touzhai Valley approximately ranges from 1 m to 2 m [11,34], and the underlying basalt bedrock gradually appears within this depth range.

Since the focus of this work is mainly on the transmission characteristics of water in the macropore network system, the gravel and matrix needed not to be distinguished and were unified into the soil matrix. Therefore, only the macropores with CT values between −1024 and −424 HU would be separated, and media with other CT values would be merged. Thus, there were only two media in the final numerical model, i.e., macropores and the matrix. In addition, pores with sizes of <1 mm were out of the scope of macropores, and their existence would increase the extra calculation amount and time, while the established model of the macropore network system only consisted of pores with diameters of >1 mm. Taking into account the complexity of the macropore system and the limited computing capacity of the computer and MODFLOW program, only the lower half of the soil column sample (20~40 cm) was modeled here (Figure 7).

The model shown in Figure 7 was generated by MIMICS. It needed to be discretized before it was imported into MODFLOW, and the process was as follows: (1) the structured grid was used to divide a grid every 1 mm, so a total of 12.5 million grids were divided; (2) grids occupied by the matrix were marked as 1, and those occupied by the macropore were marked as 2. In this way, a binary dot model file marked with 1 and 2 was obtained. By adding the header file commands and dot coordinates specified by MODFLOW, it was converted into a numerical model that could be identified by MODFLOW (Figure 8). The model was extended by 2.5 cm on both sides in the X and Y directions so that abnormal results could be avoided. Finally, the size of the MODFLOW model was a × a × h = 30 × 30 × 20 cm.

3.2. Hydraulic Conductivity and Boundaries

In MODFLOW, macropores were treated as being comprised of very coarse porous media and assigned a hydraulic conductivity (K_S), such that they would essentially behave like hollow volumes [35]. The K_S of soil matrix had been obtained through the double-ring permeability test in Wang et al. [11], but that of macropores is still uncertain.

The range of K_S for macropores in previous studies varied greatly (

Table 1. K_S values of macropores in some previous studies.

KS of Macropores(cm/s)	References
0.667	[39]
0.042	[40]
0.0042	[17]
0.061	[24]
0.126
0.167
0.046	[21]
0.023	[41]
50~70	[38]

), which was owing to the different types of soil. What is more, some scholars directly regarded macropores as conduits and believed that the macropore flow had been transformed into pipe flow [36,37,38].

Meanwhile, some scholars did not specify the K_S of macropores, but rather they investigated the ratio relationship between the matrix and macropores. For example, Elçi et al. [42] simulated the lateral macropore flow in the forested riparian wetland using MODFLOW, and the results showed that the ratio of matrix permeability to matrix plus macropore permeability was approximately 1/150; Yu et al. [43] pointed out that the K_S of macropores is 100 times that of matrix in their study. These two studies, especially the former, are similar to this paper. In this study, therefore, the average soil mass K_S acquired by Wang et al. [11] is assigned to the matrix (0.01277 cm/s), and the macropore K_S is obtained by expanding the former by 150 (1.92 cm/s) (

Table 2. K_S values of media in the MODFLOW model. The K_S of soil mass in P-1# and P-2# were from Wang et al. [11].

Experiment Sites	Depth (cm)	KS (cm/s)
		Soil Mass	Average of Soil Mass	Macropores
P-1#	0	0.0311	0.01277	1.92
	10	0.0115
	60	0.0177
P-2#	0	0.00774
	10	0.00358
	60	0.00502

).

There are two types of simulation conditions: steady seepage and transient seepage. In the steady seepage, the top and bottom surfaces of the model are constant head boundaries, with the height of H_top = 0.2 m and H_bottom = 0, separately; other surfaces are defaulted as no-flux boundaries. In the transient seepage, the initial head (IH) of the top surface is 0.40 m, consistent with the rainfall intensity of 50 mm/h (sprayed for 4 h) in the in situ dye experiment in Wang et al. [11]; the bottom surface is the infinite drain boundary, with the drainage rate of 10¹⁰ m³/day. The simulation time is T = 1 day, which is divided into 24 time-steps. Finally, all the simulation results are extracted and imported into Tecplot version 10.0 software for post-processing.

All model basic parameters are summarized in

Table 3. Overview of the model’s basic parameters.

Model Size (cm)	Boundary Condition		Hydraulic Conductivity (cm/s)
	Steady Seepage	Transient Seepage	Matrix	Macropore
30 × 30 × 20	Htop = 0.2 m Hbottom = 0	Top: IH = 0.4 m Bottom: drainage rate = 1010 m3/day	0.01277	1.92

.

4. Simulation Results

4.1. Transmission Characteristics of the Macropore Flow in the Steady Seepage Simulation

Figure 9 shows the Z-velocity distribution of macropore flow in the steady seepage field. When V_z > 0.002 m/s (Figure 9a), the displayed velocity vector occupies most of the space of the entire model. It contains some relatively regular shapes (e.g., a zone near the zero in Figure 9a), which represent the matrix flow of the soil. With the increase in the lower limit of velocity, the regular zone is further reduced (Figure 9b,c). When V_z > 0.02 m/s (Figure 9d), the regular zone basically disappears. Whereafter, if the lower limit of velocity keeps rising, the display of velocity vectors in the macropores is affected (Figure 9e,f). Hence, the velocity dividing point of matrix flow and macropore flow should be 0.02 m/s. In addition, the maximum velocity of macropore flow nearly reaches 0.3 m/s, which is very similar to the result in Weiler [38]. In this case, it is not impossible that the macropore flow could transform into pipe flow. It also indicates that macropores in vegetated slope soils play a significant role in promoting rainfall infiltration. Macropores in the soil are interconnected to form a large-scale network system that can absorb all rainfall without generating surface runoff. Specifically, this reveals that no overland flow was observed in the field dye tracer infiltration experiments in Wang et al. [11], although the rainfall intensity of 50 mm/h had far exceeded the level of heavy torrential rain.

In order to explore the velocity distribution of macropore flow in detail, some seepage fields on profiles perpendicular to the Y-axis are shown in Figure 10. Only from the presented individual profiles can we see that the matrix flow velocity in the soil is basically below 0.02 m/s and the flow velocity in isolated macropores is generally between 0.02 and 0.08 m/s. However, the flow velocity in macropores shows good connectivity. The local enlargement on each profile shows that the maximum velocity in the well-connected macropores reaches 0.23~0.30 m/s, which is approximately 10 times that in the matrix and much higher than that in isolated macropores. Meanwhile, in every single macropore, the maximum velocity appears at the central axis and gradually decreases during the transition to the macropore edge. As a result, the flow velocity distribution is mainly controlled by the connectivity of the macropores.

Another interesting phenomenon is that the macropore flow velocity is more influenced by the morphology of macropores. With the constant change in the macropore diameter, the flow will reach the maximum velocity at the throat along the extension direction of macropores. This is attributed to the smaller size of the throat compared to that of the macropore, and the infiltration pressure generated by the flow is much greater than the resistance of the throat wall, prompting the water to pass through the throat at a higher velocity. Hence, the size of the throat leads to the difference in microscopic pore structure, which thereupon determines the difference in fluid flow.

4.2. Transmission Characteristics of the Macropore Flow in the Transient Seepage Simulation

Figure 11 shows the results of the macropore flow velocity in the Z-direction at time-steps 1, 2, 3, 4, 9, 10, 11, 12, and 24. As with the steady seepage simulation, the separation of the macropore flow and matrix flow is performed by setting different lower limits of velocity. The lower velocity limit of macropore flow for all time-steps of the model is 0.02 m/s, which is consistent with that of the steady seepage condition. In the beginning, essentially all of the macropores in the model are filled with water flow (Figure 11a–f). With the elapsing of calculation time, the space occupied by macropore flow slightly decreases (Figure 11g–i). Specifically, the distribution of macropore flow at the last time step (time-step 24), compared with the early few time-steps, obviously becomes sparse. Meanwhile, the maximum velocity of the flow in the localized macropore reaches 0.623 m/s at time-step 1 (Figure 11a), 0.574 m/s at time-step 2 (Figure 11b), and 0.496 m/s at time-step 3 (Figure 11c). However, at time-step 24 (Figure 11i), the maximum velocity was only 0.207 m/s. The maximum velocity of macropore flow seems to decrease gradually with time elapsed. Figure 12 is the Z-velocity of flow on the profile of Y = 0.15 m at different time-steps and shows the same characteristics coincidentally.

Regarding uncovering the variation of macropore flow velocity with time elapsing, the maximum velocities of macropore flow at all time-steps on the profiles shown in Figure 10 are summarized in Figure 13. Throughout the entire simulation period, the maximum velocities in the same macropore on all profiles decrease with time unevenly, showing the trend of a steep drop at the outset, a slow decrease, and a gradual leveling off.

5. Discussion

5.1. Variation of Macropore Flow Velocity in the Transient Seepage Simulation

According to Darcy’s law, the flow velocity is proportional to the hydraulic gradient. The equations are as follows:

$$v=K\cdot J$$

(1)

$$J={\displaystyle \frac{H_1-H_2}{l}}$$

(2)

where v is the flow velocity, K is the hydraulic conductivity, J is the hydraulic gradient, H₁ and H₂ are the water heads at the start and end points, and l is the length of the seepage path.

In the case of transient seepage simulation, the K_S (corresponding to K) and H_bottom (corresponding to H₂) do not change with time, while v, IH (corresponding to H₁), and h (corresponding to l) are all functions of time (t). Then, Equations (1) and (2) can be rewritten as

$$v\left(t\right)=K_S\cdot {\displaystyle \frac{IH\left(t\right)-H_{\mathrm{bottom}}}{h\left(t\right)}}$$

(3)

where H_bottom = 0 in the model, and then Equation (3) is

$$v\left(t\right)=K_S\cdot {\displaystyle \frac{IH\left(t\right)}{h\left(t\right)}}$$

(4)

Therefore, v(t) is directly proportional to IH(t) and inversely proportional to h(t). Throughout the simulation time period, h(t) barely changed. However, the initial head is constantly decreasing since the bottom of the model is an infinite drain boundary, thus the macropore flow velocity in the model decreases with time elapsed.

In addition, the maximum velocity in the transient seepage simulation was 0.623 m/s at the beginning (Figure 11a) and generally decreased to 0.207 m/s at the end (Figure 11i), which were both distributed in macropores with diameters of >1 cm. Coincidentally, Weiler [38] also pointed out that macropore flow often presented high flow velocities with the order of 50~70 cm/s. In this case, the macropore flow has indeed been transformed into pipe flow. The strong water-transmitting capacity of macropores can transfer rainwater from the surface to deep in the soil and reach the soil–bedrock interface in a short time, prompting the groundwater to exhibit a rapid response behavior to the rainfall.

5.2. Implications of the Rapid Response Behavior of Groundwater to Rainfall for the Stability of Vegetated Slopes

Most landslides are the result of the time-dependent deformation of soil masses caused by changes in the stress environment. In other words, slopes generally undergo a long period of evolution before the significant displacement occurs, which is also called the landslide preconditioning process. The movement of soil masses takes place in a complex environment near the slope surface, in which the groundwater should be the protagonist. Groundwater transformed by rainfall is the most important trigger of slope soil creep, landslide preconditioning, and failure [44,45], while vegetation largely determines its path, degree, and scale in preparing slopes for failure.

The development of vegetation provides macropore channels for the downward transmission of rainwater. As plants grow up, root–soil interstices parallel to the root interface will be formed due to the uncoordinated deformation between root and soil mass; the pressure generated by root growth will cause the expansion of the overlying soil mass, forming soil cracks; organic matter produced by the vegetation metabolism entices earth animals to burrow deep into the soil, facilitating the development of worm pores; and after dying, the decomposed roots will leave tubular spaces in different scales, called rotten root channels. Since these types of macropores are often covered by the upper soils composed of semi-decaying branches and protected by fresh litter, they have a considerable degree of stability. During rainfall, the preferential flow [46,47] with a complex flow pattern appears in these stable macropores, through which the duration and amount of the groundwater recharge process will be accelerated and increased. In contrast, the macropore system in the unvegetated slope soil will gradually disappear soon after rainfall due to the lack of effective protection.

The immediate impact of vegetation on groundwater recharge in slopes is the change in soil infiltration behavior and surface runoff characteristics. As shown by the results presented in the seepage simulation (Figure 9, Figure 10, Figure 11 and Figure 12), the flow rate in macropores, especially worm pores, can be very high compared to the rate in the soil matrix. Even for relatively small pores, the flow rate seems to always be higher than the rainfall intensity [48] (Figure 10 and Figure 12). This implies that the slope surface runoff will become so weak that it could be neglected [49]. That is to say, except for loss by intercepting, rainfall water will all infiltrate and recharge the slope groundwater. Then, during an independent rainfall event, the water level in the slope aquifer will be raised significantly, causing rapid saturation in areas near the toe of the slope or stream. Meanwhile, water from these saturated areas will in turn recharge the streams as surface runoff, promoting a significant drop in groundwater levels before the next independent rainfall event, making room for the next recharge of groundwater. In this way, the dynamic pattern of frequent fluctuation of the seepage face is formed.

Frequent fluctuations in the groundwater table in slopes lead to changes in the soil structure. The rise and fall of the groundwater table puts the slope soils in a continuous process of leaching, which directly carries away the soil particles and leads to an increase in the size of the large pore space. The frequent fluctuation of the groundwater level in the slope has caused changes in the soil structure. The rising and falling of the groundwater level subjects the slope soil to continuous leaching, which directly carries away soil particles and leads to an increase in the size of the macropore space. Additionally, the acidic environment of vegetated slope soils will promote the dissolution of mineral particles in the soil, increasing the erosive effect of infiltrating rainwater on the slope soil and exacerbating the expansion of macropore size. As a result, the dual effects of leaching at the physical and chemical levels further enlarge the scale of the macropore network system.

According to the above, the initiation, development, and sustenance [37] of the macropore network system in the vegetated slope soils significantly optimizes the recharge environment of groundwater and shortens the lag time of the peak discharge response to rainfall. At this time, only when there is prolonged and high-intensity rainfall will the slope stability be threatened. Moreover, under the conditions of soil creep, groundwater infiltration, and their interactions, the gradual decline in the comprehensive quality and overall strength of the slope will occur due to the expansion of the macropore network system. This will increase the impact of groundwater on the stress field and provide a foundation for large-scale slope failure under rainfall and earthquakes.

6. Conclusions

To elucidate the facilitative effect of macropores on rainfall infiltration in vegetated slope soils, in this paper, a series of investigations were undertaken, leading to the following principal conclusions:

(1)

The distribution pattern of macropores in the soil is not coordinated but is rather in the direction of horizontal or oblique extension into the depth. The shape of the macropores contains not only a dominant proportion of circular tubes (mainly biological macropores, decreasing with the depth) but also a small proportion of flakes (mainly structural macropores). Overall, the soil macro-porosity decreases with increasing depth.

(2)

The numerical simulation results showed that the velocity of macropore flow was greater than 0.02 m/s, which was much larger than that of flow in the matrix, and the maximum flow velocity in some macropores with a diameter of >1 cm was 0.2~0.3 m/s. The better the connectivity of the macropores was, the faster the flow rate became. In every single macropore, the velocity reached the maximum at the center perpendicular to the extension and at the throat along the extension. In addition, the transient seepage simulation displayed that the transmission velocity of macropore flow decreased with time elapsed, which was attributed to the gradual decrease in the initial head over time.

(3)

Groundwater should be the primary factor responsible for slope failure. Due to the development of vegetation, a large number of macropores, which are characterized by different types but a stable existence, are formed in the slope soil, altering the infiltration behavior of the soil. Rainwater can quickly pass through the soil profile in the form of preferential flow or pipe flow, which will not only accelerate and increase the recharge process and amount of groundwater but also shorten the lag time of the peak discharge response to rainfall. This process can free up space for the next recharge, which is beneficial to slope stability. However, it also reduces the overall quality of the soil, laying the foundation for slope failure.

The aforementioned conclusions enhance our understanding of the roles played by rainfall, vegetation, soil bodies, and groundwater in landslide occurrences, which facilitates the development of targeted disaster mitigation and prevention strategies, and, thereby, improves the stability of slopes in landslide-prone mountainous urban areas. Consequently, this research significantly contributes to the construction of safer and more sustainable communities.

Moreover, the insights gained from this study promote sustainable land use practices and the establishment of healthy ecosystems, ensuring both ecological integrity and economic benefits. Such advancements align with broader sustainability goals, including the United Nations Sustainable Development Goals, particularly those pertaining to life on land, sustainable cities and communities, and climate action. This not only safeguards human lives and assets but also preserves ecological health for future generations.