Introduction

The Tibetan Plateau and its nearby mountainous areas, with frequent seismic activity, extreme rainfall conditions, deep-cut topography, and complex geological conditions, are highly prone to geological hazards, with a wide distribution area, many types, high frequency, and large scale, posing a serious threat to settlements, reservoirs, telecommunication routes, railways, and other infrastructures. The factors that cause slope deformation include geological structure, tectonic movements, seismic events, rainfall, long-term weathering, and human activities1,2,3,4. The main causes of the sudden acceleration of deformation rates on creeping slopes may be earthquakes, rainfall5, and groundwater6.

Usually, rainfall plays a crucial role in triggering landslides7,8,9,10. The variation of pore water pressure in the shear zone during the deformation process of deep-seated creeping slopes can cause sudden or delayed acceleration of the deformation rate and catastrophic damage to slopes11,12,13,14,15,16,17. Agliardi, et al.18 reported that the increase in pore water pressure in the shear zone causes an increase in the deformation rate and gradually develops into a sustained steady-rate creep, ultimately leading to failure. Zêzere, et al.19 and Scaringi20 indicate that continuous precipitation may reduce the strength of the soil, leading to deep displacement along the shear zone. Most studies focus on transient effects during periods of heavy precipitation. However, rainfall can cause strong interactions between landslide materials and water in the long term, leading to significant uncertainty in landslide risk. Therefore, it’s necessary to study the impact of multi-stage rainfall on the deformation of deep-seated landslides21.

The influence of historical rainfall on landslides is mainly manifested in the deformation and intensity change of shear zones. The occurrence of catastrophic damage to slopes caused by a certain earthquake or rainfall is highly dependent on pre-existing damage22. In real situations, landslides may not provide immediate feedback to rainfall21. The sudden deformation of the slope mainly involves the progressive failure of rocks and the accumulation of damage along the shear zone. As the shear zone gradually forms, the creep deformation is increasingly influenced by hydro-mechanical behavior14,23,24. However, for giant rock landslides that are mainly forced by the accumulation of pore pressure during intermittent or long-term acceleration periods, hydraulic coupling associated with short-term pore pressure development is the main driver of sliding behavior over time18.

Here, samples are collected from the basal shear zone of the landslide., and innovative laboratory experiments are conducted to explore the correlation between multi-stage rainfall and the deformation of a deep-seated rock landslide, the Tizicao landslide (TZC landslide), in Maoxian, Sichuan Province, China. The experiment focuses on the time-dependent response of the landslide to rainfall events, achieving indoor reproduction of the deformation evolution process of slopes under multi-stage rainfall.

The Tizicao landslide characteristics

The study area is located in the Longmenshan area at the northwest margin of the Sichuan Basin (Fig. 1). The Longmenshan thrust fault belt experiences ongoing uplift and deformation, and the average uplift rate is 300–400 m/Ma and the downcutting rate is 1,190 m/Ma. The geological environment is dominated by three parallel northeast-southwest trending thrust faults: the WC-MX Fault, the PZ-GX Fault, and the YX-BC Fault. And the nearest fault to the TZC landslide is the Shidaguan Fault, which is 2 km away from the landslide. The structural system is part of a high tectonic activity zone25. Since the twentieth century, 6 strong earthquakes with magnitudes greater than 5.5 have occurred within a 200 km radius of the Tizicao landslide, with the largest being the 8.0 magnitude Wenchuan earthquake in 200826.

Fig. 1
figure 1

Regional structural map of the study area, generated using ArcMap version 10.7 (https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview).

Full size image

The TZC landslide is situated on the right bank of the Minjiang River in Tizicao Village, Maoxian County. Figure 2a shows the three-dimensional model of the TZC landslide. The TZC landslide has a generally chair-shaped topography. The landslide is approximately 560 m in length, 600 m in width, and 1388.2 × 104 m3 in volume. Notably, the TZC landslide poses a serious threat to the national highway G213. For the safety of G213 and surrounding residents, it is necessary to monitor the landslide, understand its triggering factors, and analyze its failure mechanism.

Fig. 2
figure 2

(a) Three-dimensional model and (b) Engineering geological plot of the TZC landslide.

Full size image

According to the topography along line I-I’ (Fig. 2b), the slope of the TZC landslide can be divided into three sections (Fig. 3). The section from the Minjiang River to an elevation of 2000 m is a straight steep slope with an average slope of 47°. The toe of the TZC landslide is located at an elevation of 2000 m, approximately 270 m higher than the Minjiang River, which has a water level elevation of about 1730 m. The toe of the TZC landslide to the rear edge of the landslide is slightly gentle, with an average slope of 25°. The section above the landslide has an average gradient of approximately 45°. There are three gullies developed on the slope (Fig. 2b), namely Old Bear Cave, Nine Turn, and Large Trough gullies. The three gullies converge to form floods during a rainstorm, but there is no water flow when there is no rain. Pore water exists in Quaternary high-permeability loose sediments and is greatly influenced by climate and atmospheric precipitation. Surface water infiltrates cracks, forming underground runoff.

Fig. 3
figure 3

Geological profile (see Fig. 2b for the location of the profile line).

Full size image

There is a small amount of precipitation in winter and spring, but heavy rainfall occurs in summer and autumn (Fig. 4). More than 90% of the rainfall in the study area is concentrated from June to October, and the deformation of the TZC landslide mainly occurred during this period. In September 2013, following three consecutive months of rainfall from May to July, the deformation of the middle leading edge of the TZC landslide initiated. Simultaneously, the houses on the slope began to deform and shift downslope. In September 2014, after experiencing the rainfall (252 mm) in June, which was the maximum amount of rainfall since the deformation of the slope occurred, a rock avalanche occurred at the leading edge of the landslide (Figs. 2b and 5a), creating a slide zone approximately 60 m wide and 200 m high. At the same time, there is a group of feather-shaped cracks on the northern boundary (Fig. 5a). From August to September 2015, the landslide experienced a significant increase in deformation after a peak rainfall of 382 mm in June. After experiencing the rainy seasons in 2016 and 2017, a displacement of more than 10 m occurred at the rear edge of the landslide in July 2017, and a large number of tensile cracks appeared on the upper part of the rock avalanche and the rear edge of the landslide (Fig. 5b). Cracks have gradually widened and deepened over time, eventually forming subsurface pathways that enhance rainfall infiltration and subsurface water accumulation in the shear zone. Afterward, the TZC landslide remained in a slow deformation state, and the maximum accumulated surface deformation of the TZC landslide within two years reached 140 mm27.

Fig. 4
figure 4

Monthly rainfall of the TZC landslide, which is modified by Zhou, et al.28.

Full size image
Fig. 5
figure 5

Crack development (a) WorldView-2 image on January 9, 2016; (b) Unmanned Aerial Im-agery on July 29, 2017.

Full size image

Topple deformation features are observed at the top (Fig. 6a,b) and toe (Fig. 6c,d) of the TZC landslide. Figure 6b shows that the rock layers were curved. The dip angle decreased from 62° to 35°. Figure 6d shows that the average dip angle of the rock layer decreased from 75° to 60°. Figure 7 shows 129 orientation data of the joints. These orientation data show that the dip gradually decreases from low to high elevations. Two joint sets are developed on the TZC landslide. J1 is nearly perpendicular to the bedrock, whereas the strike of J2 is nearly parallel to the valley and tends to be outside the slope.

Fig. 6
figure 6

Photos showing the toppling deformation of the SDG slope (positions of these photos are shown in Fig. 2b).

Full size image
Fig. 7
figure 7

Contoured pole plot of stratification joints and bedding planes (Lower hemisphere, equal angle stereonets).

Full size image

We conducted 9 drills on the TZC landslide from August to September 2017 (Fig. 2b), and no stable groundwater level was detected through drilling. The drilling revealed six types of lithology (Fig. 8a), including loose deposits, moderately weathered rock mass, fragmented rock mass, highly weathered phyllite, soil in the sliding zone, and phyllite bedrock. The soil in the sliding zone is composed of silty clay with angular gravel and fragmented stones, mostly in a hard plastic shape. The thickness of the soil in the sliding zone ranges from 1.2 to 3 m, with localized areas exhibiting greater thickness, such as at borehole B09, where the thickness reaches 15.8 m. It also exhibits stratification, characterized by alternating layers of yellow–brown silty clay and gray silty clay. The soil in the sliding zone is relatively dense (Fig. 8g,h), and the drilling cores are generally well-preserved, appearing as long cylindrical columns (Fig. 8b).

Fig. 8
figure 8

(a) Borehole histogram, lithofacies Characteristics, RQD, (b) a photo of borehole cores: soil in the sliding zone and phyllite, SEM results of (c, d) B08 and (e) B09, the sliding zone characteristics revealed by the trail trench (f), the wells TJ1 (g), and TJ2 (h).

Full size image

At the same time, we conducted scanning electron microscopy (SEM) to observe the microstructure of the sliding zone material at the same depth in boreholes B08 and B09, to reveal the movement structural characteristics of the landslide. From the SEM results, the rock sample of B08 located on the north side of the landslide has obvious scratches (Fig. 8c,d). TC1, located at the rear edge of the landslide, also exposes distinct characteristics of the sliding zone (Fig. 8f). The rock sample of B09 located on the south side of the landslide also showed scratches, but the scratches were arc-shaped and the soil was relatively intact without fragmentation characteristics (Fig. 8e). The formation of these scratches mainly came from slow creep friction. The north and south sides of the TZC landslide have different motion characteristics in the sliding zone.

The cumulative distribution curve of grain size composition of the sliding zone material was obtained by accumulating the percentage content of different particle groups in the sliding zone material of the TZC landslide (Fig. 9). The coefficient of uniformity (Cu) is greater than 10, the coefficient of curvature (Cc) is less than 1, and the grain size distribution of the sliding zone material of the TZC landslide is uneven. The diffraction pattern of the sliding zone soil was obtained from X-ray diffraction (XRD) (Fig. 10). The XRD result shows that sliding zone material is mainly composed of illite, chlorites, quartz, and plagioclase, and the content of clay minerals is the highest, accounting for 58%.

Fig. 9
figure 9

The cumulative distribution curve of grain size composition of the sliding zone material.

Full size image
Fig. 10
figure 10

XRD result.

Full size image

In 2019 and 2020, we conducted electrical resistivity tomography (ERT) measurements on the same longitudinal profile (A-A’ in Fig. 2b) on the landslide using the Wenner device. The electrode spacing was 10 m, the profile length was 800 m, and the maximum detection depth reached 144 m. During electrical measurements, the following parameters were set: electrode grounding resistance was kept below 2 kΩ·m. The obtained data from April 2019 was used as the initial model for time-lapse electrical resistivity tomography calculated using Res2DInv (version 3.54.44, and URL link: https://www.aarhusgeosoftware.dk/download-resxdinv) program based on the least-squares method29. The number of iterations was set to 5, and the RMS error was maintained at 10%. It can be clearly seen that resistivity decrease and resistivity increase are present in intervals in the shallow part (0–15 m) of the A-A’ profile (Fig. 11). Resistivity increases by more than 100% along the downward extension path of cracks and by 50–80% in parts of the basement. The landslide generates relative shear displacement along the base surface, exacerbating the development of surface cracks and deep fractures. A decrease in resistivity can be observed near the base surface with a reduction amplitude exceeding 45%. This represents the infiltration of rainwater into the basement surface through the near-vertical fissure zone and the accumulation of water at the base of the bedrock fissures, creating a water-rich zone at the concave basement surface. Spearman correlation analysis was conducted using GNSS monitoring data and rainfall records, showing a significant correlation between displacement and rainfall27. To further investigate the mechanism by which rainfall influences the deformation of the Tazichong landslide, this study carried out rainfall-step creep experiments to explore the time-dependent deformation behavior under rainfall conditions.

Fig. 11
figure 11

Time-lapse electrical resistivity tomography model of TZC landslide (A-A’).

Full size image

Method

We study the time-dependent response of the basal shear zone of landslides to short-term changes in pore water pressure, as well as the gradual deformation evolution process of slopes under multiple periods of rainfall. The shear zone material found between 41.1 and 48.2 m in borehole B08 was sealed and used in the rainfall-step creep experiments. We design a water amount intensifier to simulate rainfall (Fig. 12), and injected the water volume designed in the sample plan directly from the top of the sample through a water flow pipe. Using the consolidation procedure of the MTS810 dynamic triaxial testing machine, rainfall-step creep experiments are carried out on the shear zone material of the TZC landslide to simulate the effect of rainfall on the shear zone by adding the design water volume to the specimen through the water amount intensifier. In order to be more in line with the geological evolution process of the shear zone of the TZC landslide, the same stress loading rate (1 kPa/s) is first applied to the same axial and confining pressures. After completion of consolidation under the equal pressure condition, the axial pressure of the specimen is increased to a preset stress (stress ratio 1:1.2) at a rate of 1 kPa/s and remains constant throughout the experiment. The creep test is started after completion of consolidation under the unequal pressure condition. In order to more accurately reveal the effect of each rainfall on the slope deformation, we waited for the deformation of the specimen to stabilize (deformation < 0.1 mm in 10 min and stable pore pressure) before applying the next rainfall. Before each rainfall application, we closed the drainage valve to prevent water from draining out immediately under a strong hydraulic head. After completing rainfall infiltration, the drainage valve was reopened. If the deformation reaches 30%, the sample can be considered as a failure and the test is terminated immediately. The rainfall-step creep experiments are conducted under the drained condition. The test procedure is shown in the Fig. 13.

Fig. 12
figure 12

Experimental apparatus.

Full size image
Fig. 13
figure 13

Experimental procedure.

Full size image

The MTS810 dynamic triaxial testing machine can apply an axial load of 0–25 kN, confining pressure of 0–2000 kPa, and axial displacement measurement of 0–50 mm. The samples for the experiments are remolded samples from the shear zone of the TZC landslide, with dry density \(\rho_{d}\) = 1.85 g/cm3, water content w = 5%, and permeability coefficient k = 1.682 × 10−5 cm/s, which is obtained through constant head permeability tests. The specimens are carefully formed into cylindrical standard samples, each with a diameter of 50 mm and a height of 100 mm. The specimen is compacted in five layers, with each layer carefully leveled to ensure uniformity throughout the specimen. And only soil particles smaller than 10 mm were used in the tests, in accordance with the requirement that the maximum particle size be less than 1/5 of the specimen diameter in triaxial creep testing.

We design the rainfall-step creep experiments to simulate the deformation evolution of the sliding zone material of the TZC landslide under different rainfall conditions. According to the rainfall classification in China, 12-h rainfall less than 5 mm is considered light rain, 5–14.9 mm is considered moderate rain, 15–29.9 mm is considered heavy rain, 30–69.9 mm is considered rainstorm, 70–139.9 mm is considered heavy rainstorm, and greater than or equal to 140 mm is considered extremely heavy rainstorm. According to the rainfall conditions on the TZC landslide, we selected moderate rain (1430.99 mm3), heavy rain (3066.41 mm3), and rainstorm (6132.81 mm3) for the rainfall-step creep experiments.

The test plan is shown in Table 1. In test 2, in order to explore the response characteristics of the strength of the sliding zone material response to multi-stage rainfall during the rainfall-step creep experiments, the specimens are subjected to 4, 8, and 12 heavy rain at three different confining pressures of 100 kPa, 200 kPa, and 300 kPa. After consolidation and stabilization, the undrained triaxial tests are carried out, with the same shear rate (0.5 mm/min) as test 1, until the axial strain of the specimen reaches 30%. Tests 3 and 4 are carried out by applying rainfall several times to achieve an axial strain of 30% in the specimen.

Table 1 Test plan.
Full size table

Result

The specimen is subjected to a triaxial consolidated undrained shear test (test 1), and the undrained peak strength (\(\tau_{f}\)) is obtained under three confining pressures of 100 kPa, 200 kPa, and 300 kPa (Table 2). According to the effective stress path (Fig. 14a), the cohesion of the sliding zone material is 11.902 kPa, and the internal friction angle is 11.214°. As the confining pressure increases (σ₃ from 100 to 300 kPa), the peak pore water pressure becomes higher (Fig. 14b).

Table 2 Peak strength values of specimens under different confining pressures.
Full size table
Fig. 14
figure 14

Friction characteristics and stress evolution. (a) Effective stress path of the triaxial consolidated undrained shear test; (b) Effective stress path diagram of the rainfall-step creep experiments.

Full size image

We also conducted triaxial tests under varying numbers of rainfall events to analyze the trends in shear strength. As the number of rainfall events increased, the internal friction angle showed no significant change, but the cohesion decreased markedly. The cohesion dropped from 12 kPa (before rainfall) to less than 4 kPa after 12 rainfall events (Fig. 15).

Fig. 15
figure 15

The change of shear strength (test 2).

Full size image

According to the ideal creep curve (Fig. 16), the creep process can be divided into three stages: primary creep, secondary creep, and tertiary creep, corresponding to time intervals T1 and T2.

Fig. 16
figure 16

The ideal creep cuvre, which is modified by Bhat, et al.30.

Full size image

Triaxial tests are conducted under conditions 2, 3, and 4 with a confining pressure of 100 kPa (Fig. 17). From Fig. 17a, as the rainfall volume increases, the number of rainfall decreases. Under condition 2, the sample still remains undamaged even after 110 rainfall. Under condition 3, the sample fails during the 16th rainfall. Under condition 4, the sample is damaged during the 7th rainfall. Under condition 2, the pore pressure increases with a sudden increase in specimen deformation, but the deformation rate tends to zero in a short period of time. The sudden deformation and creep deformation of the specimen due to a single rainfall is small, so the TZC landslide tends to a stable state in the long term under condition 2. However, the effect of the sudden increase in pore pressure on the specimen is cumulative and leads to the failure of the specimen after multi-stage rainfall. Under conditions 3 and 4, both the sudden deformation and creep deformation increase with the increase in rainfall frequency (Fig. 17b,c). Under condition 4, the specimen underwent rapid deformation within a short period, with axial strain quickly exceeding 5%, followed by failure. The curve clearly shows the process of initial primary creep (T1), a brief secondary creep phase (T2), and then a rapid transition into tertiary creep, ultimately resulting in failure. Under condition 3, the deformation curve exhibits a similar trend, but with slightly longer durations in both primary and secondary creep stages, and the overall deformation development is slower than that under Condition 4. The specimen eventually entered tertiary creep and failed at around 30,000 s. In contrast, under condition 2, the axial strain increased significantly more slowly. The specimen remained in the primary and an extended secondary creep phase, with no signs of tertiary creep, and no failure occurred.

Fig. 17
figure 17

Test 3: (a) Strain–time plot under different rainfall conditions with confining pressure of 100 kPa; (b) Sudden deformation-time plot, creep deformation-time plot under condition 3; (c) Sudden deformation-time plot, creep deformation-time plot under condition 4.

Full size image

Triaxial tests are conducted under conditions 3 with a confining pressure of 100 kPa, 200 kPa, and 300 kPa (Fig. 18). From Fig. 18, it can be seen that under three confining pressures, the sliding zone material rapidly deforms under the pore water pressure, but the deformation rate quickly stabilizes until it reaches 0. The greater the confining pressure, the greater the deformation of the sliding zone material, but it can withstand a greater number of rainfalls. After the deformation of the sample exceeds 15 mm, the sample usually undergoes rapid deformation during the last rainfall stage until it destroys. Under confining pressures of 200 and 300 kPa, unlike the confining pressure of 100 kPa, the creep displacement and sudden displacement first increase rapidly with the increase of rainfall frequency, then the deformation rate slows down and tends to stabilize (Fig. 18b,c). Under the confining pressure of 100 kPa, the specimen quickly enters the secondary creep (T2) without a clearly defined primary creep (T1). The strain rate remains steady but continues to increase, and the specimen abruptly enters the tertiary creep at around 20,000 s, showing signs of sudden instability and failure. Under 200 kPa, the primary creep (T1) lasts slightly longer, and the secondary creep stage (T2) is more distinct compared to the 100 kPa condition. The strain accumulates more gradually, and failure does not occur until approximately 30,000 s. In contrast, under a higher confining pressure of 300 kPa, the specimen exhibits a shorter primary creep stage (T1) than at 100 kPa, while the secondary creep stage (T2) lasts the longest among the three. The overall strain rate is the slowest, and the deformation develops steadily. The specimen enters the tertiary creep and fails only after about 45,000 s.

Fig. 18
figure 18

Test 4: (a) Strain–time plot under condition 3 with different confining pressures; (b) Sudden deformation-time plot, creep deformation-time plot with confining pressure of 200 kPa; (c) Sudden deformation-time plot, creep deformation-time plot with confining pressure of 300 kPa.

Full size image

Each time rainfall is applied, the pore water pressure significantly increases. According to the principle of effective stress, the effective stress decreases accordingly, and the axial strain of the specimen suddenly increases. The strain–time curve shows a “stepped” upward climbing, and rainfall has a significant impact on the strength of the sliding zone material, as shown in Fig. 19. In the rainfall-step creep experiments, the specimen exhibits certain attenuation creep characteristics. After rainfall, the specimen deforms rapidly, the displacement increases rapidly. In a short period of time, the deformation rate reaches its maximum value, after which it becomes smaller and smaller until it tends to a constant value. The duration of the acceleration stage is much smaller than that of the creep stage. At the same time, we statistically analyzed the time for the sample to reach the peak velocity (\({\text{TV}}_{p}\)) after each rainfall in tests 3 and 4, and obtained a consistent pattern. As the number of rainfall increases, \({\text{TV}}_{p}\) decreases (Fig. 20).

Fig. 19
figure 19

Time series plot of strain, pore water pressure, and displacement rate under condition 3 with the confining pressure of 100 kPa.

Full size image
Fig. 20
figure 20

Plot of the time to reach peak velocity (\({\text{TV}}_{p}\)) versus rainfall frequency.

Full size image

Discussion

The long-term evolution of the mechanical properties of the shear zone materials is closely related to their particle size composition, mineralogical characteristics, and chemical composition31,32,33. The XRD result (Fig. 10) reveals that the shear zone materials is composed of minerals such as illite, chlorites, quartz, and plagioclase. Areas with high clay mineral content are prone to landslides30. This mineral has a very high specific surface area, cation exchange capacity and affinity for water. When water infiltration, water molecules migrate into the interlayers of the clay mineral structure to form a polarized water molecule layer. The polarized water molecule layer weakens the connection between clay minerals, thereby significantly reducing the shear strength of the soil34,35,36,37.

The TZC landslide is in a slow deformation process. New cracks, holes and empty space continue to appear on the surface and inside the slope (Fig. 21a). Some cracks and holes are filled with argillaceous materials (Fig. 21c). And these newly generated surface cracks and deep fractures provide potential flow paths for rainfall infiltration, allowing more precipitation to flow towards the basal interface (Fig. 21a). For example, in 2020, water enrichment had already formed near the basal interface of the TZC landslide (Fig. 11). The expansive and softened clay minerals not only weaken their own cohesion and strength but also play a lubricating role in the phyllite, significantly reducing the shear strength of the phyllite. According to the triaxial tests, a clear decrease in the cohesion of shear zone materials with the increasing number of rainfall events (Fig. 17). At the same time, clay minerals will lead to low permeability. According to the variable head permeability test, the permeability of the shear zone materials is 1.682 × 10-5 cm/s. During rainfall, water is difficult to drain, and pore water pressure gradually accumulates, further weakening the effective stress and promoting deformation development. According to the experimental results, under a stress condition of 100 kPa, when the rainfall is small, the strain rate is the slowest and the landslide remains stable (Fig. 16), because the infiltration amount is small and can be discharged in time. However, as rainfall increases, the chlorite becomes saturated, leading to an expansion of the interlayer spacing. This enhances the lubricating effect within the phyllite, causing a rapid decline in cohesion. Consequently, the duration of the secondary creep stage shortens, and the landslide quickly enters the tertiary creep.

Fig. 21
figure 21

Formation and evolution process of the TZC landslide. (a) Cracks formation and precipitation; (b) the formation of the sliding surface; (c) flexural rock masses, hollow structures, and argillaceous fillings at the slope foot.

Full size image

After a single rainfall, the shear zone material may undergo reconsolidation and undergo certain strength recovery38,39,40,41. Reconsolidation can indeed enhance the strength and density of soil, but its recovery effect is significantly affected by factors such as soil particle size distribution, clay mineral structure characteristics and chemical composition8,39,42. In the case of high clay mineral content or high plasticity, reconsolidation may not be sufficient to completely offset the structural softening and strength reduction caused by rainfall. Under multiple rainfall, the structure of the shear zone material continues to weaken, leading to the gradual accumulation of pore water pressure and a significant reduction in effective stress. The landslide becomes increasingly sensitive to rainfall-induced deformation. Experimental results indicate that as the number of rainfall increases, the time to reach the peak velocity shortens (Fig. 20), suggesting a lowered threshold for sliding initiation and a higher likelihood of entering the tertiary creep. Notably, when the soil strength has not fully recovered, even lower pore pressure from renewed rainfall can trigger failure (Fig. 14b).

Therefore, under the influence of multi-stage rainfall, the landslide body undergoes a cyclic process of “softening–partial recovery–resoftening–instability”. Once the cumulative softening effect induced by rainfall exceeds the strength threshold recoverable through reconsolidation, the landslide may undergo rapid sliding within a very short time, resulting in catastrophic failure (Fig. 21). So we need to conduct long-term monitoring of the sliding zone of the TZC landslide, understand the long-term deformation evolution process, enabling timely evacuation of residents and the implementation of emergency response measures when necessary.

Conclusion

We reproduce the deformation evolution process of the TZC landslide under multi-stage rainfall through rainfall-step creep experiments and study the weakening effect of multi-stage rainfall on landslide fatigue. The results show that each rainfall causes a rapid increase in pore water pressure, leading to an acceleration of deformation in the short term, followed by secondary creep. As the number of rainfall events increases, the duration of the secondary creep is significantly shortened and failure occurs more quickly. For the same amount of rainfall, lower confining pressures resulted in slower deformation development, but damage occurred earlier. The experimental results effectively reproduce the time-varying deformation process of the TZC landslide and provide physical insights into its actual behavior. However, this study was limited to laboratory-scale tests. In the future, a combination of field monitoring and numerical modeling should be used to validate the experimental results and to better understand the quantitative relationship between rainfall and intensity decay. This will help improve the accuracy of landslide prediction and warning.