Geotechnical hazard to power facilities from the Hongyanzi landslide in Hanyuan County, China: an integrated InSAR monitoring and seepage simulation study

Abstract

Introduction:

Landslide monitoring in steep and vegetated mountainous areas is frequently limited by geometric distortions and temporal decorrelation in SAR observations due to geometric distortions and severe temporal decorrelation, which can significantly reduce the effectiveness of conventional InSAR analyses. In addition, understanding the hydrological mechanisms controlling landslide activity is difficult because numerical simulation of landslide seepage processes remains challenging because of uncertainties in boundary conditions and hydrological parameters, particularly in reservoir-affected slopes.

Methods:

To address these challenges, this study focuses on the Hongyanzi landslide in Hanyuan County, Sichuan Province, China, and explores a combined remote sensing and numerical modeling strategy. Multi-temporal Sentinel-1 ascending and descending SAR images are processed using the Intermittent Small Baseline Subset (ISBAS) approach to improve deformation coverage in low-coherence areas. To characterize zones with larger surface displacements, ALOS-2 pixel offset tracking is ho applied. Furthermore, a finite-element seepage model is established to investigate pore-pressure variations induced by changes in reservoir water levels.

Results:

The results reveal line-of-sight deformation rates of up to approximately 50 mm/yr, with broadly consistent spatial patterns derived from different viewing geometries. The seepage modeling suggests that reservoir water-level fluctuations play an important role in controlling pore-pressure evolution and slope stability, although some differences remain between simulated and observed hydraulic heads.

Discussion:

Overall, the proposed integrated approach provides improved deformation characterization and insight into the hydro-mechanical processes governing reservoir-related landslides, demonstrating its practical potential for landslide hazard assessment and early-warning support in mountainous regions influenced by reservoir operations, particularly where the safety of downstream infrastructure, such as power transmission facilities, is of concern.

1 Introduction

Landslides in mountainous regions represent a persistent threat to the operation and safety of power infrastructure, including transmission corridors, substations, and hydropower facilities. The Hongyanzi landslide (Vaka et al., 2024), located in Hanyuan County of Sichuan Province, China, lies within a region characterized by a high density of energy-related infrastructure. Infrastructure located in complex geotechnical environments is particularly sensitive to ground deformation processes that can affect structural stability and operational reliability (Ahmad et al., 2019; Ahmad et al., 2026). Previous studies have demonstrated that subsurface thermal and hydro-mechanical processes can significantly influence the stability and operational reliability of energy infrastructure systems (Ahmad et al., 2021; Ahmad et al., 2025a). Recent studies have also explored data-driven approaches to assess infrastructure degradation processes in complex soil environments (Ahmad et al., 2025b). Instability of this slope has the potential to disrupt electricity transmission, damage key equipment, and, under extreme conditions, compromise downstream hydropower systems (Xu and Zhou, 2025). For this reason, maintaining continuous and reliable landslide monitoring is a critical requirement for the long-term security and resilience of regional power networks (Huang et al., 2024).

Interferometric Synthetic Aperture Radar (InSAR) is now routinely applied to detect (Solari et al., 2020) and quantify surface deformation associated with landslide activity (Aswathi et al., 2022). Compared with conventional ground-based monitoring approaches, such as Global Navigation Satellite System (GNSS) measurements, total station surveys, or in situ sensors, InSAR provides several important advantages. First, satellite SAR observations enable wide-area and spatially continuous measurements of ground deformation, which is particularly valuable for large or inaccessible mountainous slopes where the installation and maintenance of ground instruments are difficult and costly. Second, InSAR can provide millimeter-level deformation sensitivity and long-term temporal coverage using regularly acquired satellite imagery, allowing the detection of slow-moving landslides that may not be captured by episodic field surveys. Third, satellite observations allow retrospective analysis of deformation evolution using archived SAR datasets, which is especially useful for evaluating long-term slope behavior prior to the occurrence of hazardous events.

In mountainous terrain, however, strong relief and dense vegetation frequently degrade SAR signal quality. These factors introduce geometric distortions and temporal decorrelation that reduce the effectiveness of conventional InSAR approaches, particularly in areas where power infrastructure is commonly deployed (Monti-Guarnieri et al., 2020). Methods such as the Intermittent Small Baseline Subset (ISBAS) have been proposed to improve deformation retrieval under low-coherence conditions (Cigna et al., 2021). Even with these advances, interpreting landslide kinematics and linking observed deformation to external drivers, such as reservoir water-level variations, remains difficult in practice (Tang et al., 2025).

In addition to deformation monitoring, the hydrogeological processes that control landslide initiation and acceleration play a key role in hazard assessment, especially for slopes influenced by reservoir operations (Zou et al., 2021). Numerical modeling approaches have been widely applied to investigate complex geomaterial behavior and to improve understanding of deformation mechanisms in geotechnical systems (Rizvi et al., 2020). Finite-element based analyses are frequently used to investigate the stability of underground structures and surrounding geomaterials under varying loading conditions. Computational modeling techniques are increasingly used to evaluate fluid and transport processes that influence infrastructure performance and system reliability (Haroon et al., 2017). Numerical seepage modeling is widely used to examine pore-pressure evolution within landslide bodies, as pore pressure directly affects effective stress and slope stability (Pelascini et al., 2022; Yu et al., 2024). However, the combined use of InSAR-derived deformation measurements and seepage simulations is still relatively limited, particularly in studies aimed at reducing risks to power infrastructure (Lau et al., 2024).

In this study, we present an integrated approach combining multi-temporal InSAR analysis (Pan and Shi, 2023) and seepage numerical simulation to investigate the deformation behavior and failure mechanisms of the Hongyanzi landslide. Unlike previous studies that typically analyze InSAR observations and numerical simulations independently, our approach explicitly links time-series deformation patterns with transient hydrological processes to better constrain landslide triggering mechanisms. Using Sentinel-1 ascending and descending SAR data processed with the ISBAS method (Valade et al., 2012), we derive high-resolution deformation time series and further decompose the LOS measurements into east–west and vertical displacement components to characterize the multidimensional deformation behavior of the slope (Cigna et al., 2021). Furthermore, we develop a finite-element model to simulate transient seepage conditions under fluctuating reservoir levels (Godt et al., 2012; Yue et al., 2024), aiming to identify the hydraulic triggers of landslide activity. By linking deformation patterns with seepage dynamics (Sun et al., 2023), this work provides a scientific basis for landslide early warning and risk management strategies focused on the protection of power infrastructure in mountainous environments.

Compared with previous studies, the main contributions of this work are threefold: (1) the integration of multi-geometry InSAR observations to resolve multidimensional deformation patterns of the landslide; (2) the coupling of satellite-derived deformation time series with transient seepage modeling to investigate hydro-mechanical triggering processes; and (3) the establishment of a physically interpretable framework that links surface deformation observations with subsurface hydrological dynamics.

The results improve the understanding of deformation processes at the Hongyanzi landslide (Huang et al., 2019) and offer a transferable methodological reference for safeguarding energy infrastructure in other regions with complex terrain and similar geological settings (Vaka et al., 2024).

2 Study area and data sources

2.1 Study area

The Hongyanzi landslide is located on the eastern bank of the Dadu River in Hanyuan County, Sichuan Province, China (Han et al., 2018), as shown in Figure 1a. The area is mountainous, with steep slopes and active geological conditions (Huang et al., 2019). It also lies within a region influenced by large hydropower projects, where human activities and natural processes interact.

FIGURE 1

The landslide area is close to important power infrastructure, including high-voltage transmission lines and the reservoir of the Pubugou Hydropower Station (Yue et al., 2024). The landslide is about 700 m wide along the riverbank and extends roughly 300 m inland, with a total area of approximately 0.21 km² (Huang et al., 2019). Field observations and previous studies indicate that slope movement is affected by both seasonal rainfall and changes in reservoir water level, as shown in Figure 1b, making this site a typical example of a high-risk landslide in a reservoir-influenced setting (Jin et al., 2009).

Deformation of the Hongyanzi landslide poses a direct risk to nearby power transmission corridors and hydropower facilities, which could result in power supply interruptions and damage to infrastructure (Dai et al., 2002). In addition, several major highways are located close to the landslide boundary, increasing the potential consequences of slope instability for regional transportation and energy networks (Petley, 2012).

2.2 Data sources

2.2.1 SAR data for deformation monitoring

To monitor surface deformation, this study utilized C-band SAR data acquired by the Sentinel-1A satellite between January 2019 and June 2022. A total of 105 ascending orbit images and 100 descending orbit images were obtained from the European Space Agency (ESA) (Torres et al., 2012). Key parameters of the dataset are summarized in Table 1. The selection of both ascending and descending orbits allows for the retrieval of two-dimensional and vertical displacement components, critical for comprehensive slope stability assessment near power facilities (Ferretti et al., 2001). The Sentinel-1 data were processed using the ISBAS method to enhance coherence in vegetated and steep terrain, thereby improving the reliability of deformation measurements in challenging landscapes where power lines often traverse (Berardino et al., 2002).

2.2.2 Geological and hydrological data

Geological parameters and subsurface characteristics were derived from a published study in Geofluids, including stratigraphic profiles, material properties, and soil-water characteristic curves. Due to identified inconsistencies in supplementary CAD profiles, the digital geological model was reconstructed based on the published reference to ensure consistency. Reservoir water-level data and local rainfall records were provided in Excel format and covered the same monitoring period as the SAR acquisitions. These data are essential for understanding water-induced slope instability mechanisms, particularly relevant for hydropower dam safety and the management of reservoir shore slopes where power infrastructure may be located (Tresoldi et al., 2019).

2.2.3 Additional SAR data for large deformation analysis

To capture potential large-magnitude movements that may exceed the detection capability of conventional InSAR, L-band ALOS-2 data acquired on four dates between 2018 and 2020 were processed using Pixel Offset Tracking (POT) (Jin et al., 2009). Key parameters of the dataset are summarized in Table 2. This technique provides complementary two-dimensional deformation measurements without phase unwrapping, offering enhanced insight into significant ground displacements that could endanger adjacent power transmission towers or access roads.

The multi-source data ensemble supports an integrated analysis of both gradual and abrupt slope deformations and their hydrogeological triggers, forming a solid basis for assessing landslide risks to regional power infrastructure (Bertolini et al., 2005).

3 Methods

3.1 InSAR-based deformation monitoring

To monitor the surface deformation of the Hongyanzi landslide, the ISBAS method was employed. This technique extends the conventional Small Baseline Subset (SBAS) approach by incorporating pixels with intermittent coherence, thereby increasing spatial coverage in vegetated and steep areas—common conditions in regions where power transmission lines and hydroelectric infrastructure are located (Bateson et al., 2015). The method is particularly suitable for long-term monitoring of slow-moving landslides that may threaten the structural foundation of power towers and access roads (Valade et al., 2012). Sentinel-1A ascending and descending orbit images were processed to derive line-of-sight (LOS) deformation time series. Furthermore, a multi-dimensional deformation analysis was conducted by combining ascending and descending datasets under the assumption of a parallel displacement field, enabling the estimation of horizontal (east-west) and vertical movements (Elmore et al., 2016). This parallel displacement field assumption implies that the displacement direction is spatially consistent within the landslide body, meaning that individual points may experience different magnitudes of displacement but their motion vectors are approximately parallel (Wright et al., 2004).

3.1.1 Interferometric phase composition

The interferometric phase for the kth interferometric pair can be expressed as:where:

is the radar wavelength (approximately 5.6 cm for Sentinel-1 C-band).

is the LOS deformation displacement between two acquisitions.

represent orbital error, atmospheric delay, residual topographic phase, and noise, respectively.

3.1.2 Time series deformation calculation

Assuming M interferograms are formed from N+1 SAR acquisitions, the system of phase equations can be written in matrix form:where:

A is an M × N design matrix defining the temporal relationships between acquisitions.

is the vector of unknown phase values (proportional to displacement) at each time epoch.

is the vector of unwrapped interferometric phases.

The system is solved using Least-Squares estimation, often stabilized via SVD due to the rank deficiency of A.

3.1.3 ISBAS enhancement

ISBAS (Figure 2) extends the conventional SBAS approach by relaxing the requirement that pixels must remain coherent over all interferograms. Instead, pixels are selected based on intermittent coherence, allowing deformation time series to be estimated using only those interferograms in which a given pixel exhibits sufficient phase quality. As a result, a larger number of pixels can be inverted, particularly in areas affected by vegetation cover or strong temporal decorrelation.

FIGURE 2

For each pixel, deformation is retrieved by solving a least-squares inversion based on an incomplete interferogram network, where the design matrix and observation vector vary depending on the availability of coherent interferograms. This strategy enables meaningful deformation signals to be recovered even when coherence is not maintained continuously over time. Such capability is particularly important for monitoring landslides and other unstable slopes near power infrastructure in mountainous regions.

3.2 Multi-dimensional deformation decomposition

Interferometric Synthetic Aperture Radar (InSAR) measures ground deformation along the radar line-of-sight (LOS) direction. Because a single LOS measurement represents only the projection of the three-dimensional displacement vector onto the radar viewing direction, it cannot independently resolve the full deformation field. To retrieve multi-dimensional surface motion, LOS observations from both ascending and descending satellite tracks were combined to estimate horizontal and vertical displacement components.

Let the three-dimensional ground displacement vector be expressed in the local east–north–up coordinate system aswhere , , and represent the east–west, north–south, and vertical deformation components, respectively.

For a given radar acquisition geometry, the LOS displacement is related to the 3-D displacement vector through the projection of the radar look vector:

Where is the radar incidence angle, and is the satellite heading angle measured clockwise from north.

Because near-polar orbit SAR systems have very limited sensitivity to motion in the north–south direction, the contribution of is typically small and poorly constrained. Therefore, following common practice in InSAR deformation studies, the north–south component was assumed negligible, allowing the deformation field to be approximated using only the east–west and vertical components.

Under this assumption, the LOS observations from ascending () and descending () tracks can be written aswhich can be expressed in matrix form:

The east–west and vertical deformation components were obtained by solving the above linear system using least-squares inversion. This approach combines the complementary viewing geometries of ascending and descending orbits to partially reconstruct the two-dimensional deformation field.

Prior to the decomposition, the ascending and descending InSAR time series were co-registered onto a common geographic grid to ensure spatial consistency. Only pixels with valid measurements in both tracks were retained for the decomposition. The resulting east–west and vertical deformation fields provide improved physical interpretation of ground motion processes compared with single-track LOS observations.

3.3 Pixel offset tracking for large deformations

POT (Figure 2) estimates 2D deformation by maximizing the cross-correlation coefficient between patches of image amplitude in master and slave SAR images (in both range and azimuth directions), bypassing phase unwrapping (Herman et al., 1999).

The sub-pixel offset , which contains deformation information, is found by fitting a function (e.g., a parabolic surface) to the correlation peak and finding its maximum. The final deformation components are derived by:where the second term in accounts for the offset due to baseline decorrelation, is the slant range, is the incidence angle, is the perpendicular baseline, is the Pulse Repetition Frequency, and is the satellite velocity.

3.4 Numerical seepage simulation

In this study, considering the scale of the landslide and computational efficiency, the mesh size for the sliding body was set to 5 m, while that for the sliding bed was set to 15 m (Vaka et al., 2024). The entire model comprises 1058 nodes and 962 elements. It is well established that the specification of boundary conditions is critical for accurately simulating the seepage field within a landslide. Previous studies have demonstrated that rainfall can induce a hydraulic gradient along the slope direction in reservoir landslides (Fan et al., 2024; Liu et al., 2025). A rapid drawdown of the reservoir water level generates significant hydrodynamic pressure, whereas a rise in water level induces an inward-directed uplift force (Fan et al., 2024; Liu et al., 2025). These processes represent key hydraulic factors influencing the seepage field and stability of reservoir landslides. Real-time water level data for the Pubugou Hydropower Station were obtained from the relevant authorities (Liu et al., 2026). Additionally, daily rainfall records were acquired from the Hanyuan Meteorological Station (ID 56376; 29°21′ N, 102°38′ E), located approximately 12 km from the Hongyanzi Landslide (Liu et al., 2026). Based on these data, three distinct hydraulic boundary conditions were considered to investigate the associated changes in pore water pressure. First, to assess the influence of rainfall alone, a unit flow boundary condition was applied to the landslide slope, with the flux value corresponding to the daily rainfall intensity. Second, to examine the combined effect of rainfall and water level drawdown, a total head boundary condition was imposed on the reservoir slope. In this case, the total head was set equal to the reservoir water level to simulate the seepage field during a drawdown from 850 m to 790 m. Third, the combined effect of rainfall and water level rise—specifically the buoyancy effect at the landslide toe—was simulated for a rise from 790 m to 850 m. At the trailing edge of the landslide, a total head boundary condition corresponding to a water level of 850 m was applied based on findings from existing studies (Vaka et al., 2024). An impervious boundary condition was assigned to the base of the landslide (Vaka et al., 2024).

The finite element method (FEM) is used to discretize the domain into elements and solve the governing equation numerically under specified initial and boundary conditions (e.g., constant head, flux, or seepage face), providing the pore water pressure distribution critical for slope stability analysis near power facilities.

4 Results and discussion

4.1 InSAR-derived deformation rates

Time-series analysis of Sentinel-1 SAR data using the ISBAS technique revealed significant movement of the Hongyanzi landslide between January 2019 and June 2022.

Figures 3a,b present the average deformation velocity maps derived from descending and ascending SAR acquisitions, respectively. The overall deformation pattern is consistent between the two viewing geometries, indicating that the InSAR-derived results are reliable. Most areas around the reservoir remain stable, with velocities ranging within ±10 mm/yr, as shown by the predominance of green pixels.

FIGURE 3

In contrast, localized deformation anomalies are clearly identified on the western bank of the reservoir. Both datasets reveal a cluster of persistent scatterers with significant negative velocities, reaching up to −30 mm/yr in the descending track and −60 mm/yr in the ascending track. These anomalies are concentrated near the monitoring points 23GP–26GP(GPS monitoring points), which are located in the central part of the slope. The deformation signals exhibit a coherent spatial distribution, suggesting the presence of a slowly accelerating landslide body.

Compared with the surrounding slopes, the deformation in this sector is much stronger and more spatially continuous, implying a higher susceptibility to mass movement. The consistency between descending and ascending tracks further confirms the reliability of the deformation detection, reducing the likelihood of atmospheric or orbital artifacts dominating the signal.

These results reveal a generally consistent motion away from the satellite along the line-of-sight direction. The ascending track appears more sensitive to north–south deformation compared with the descending track, providing valuable information for evaluating potential impacts on nearby power transmission corridors, where the direction of movement can affect infrastructure stability.

The spatial distribution of deformation is notably heterogeneous across the landslide body. Areas with low coherence, often corresponding to steep or vegetated slopes, were excluded during processing. In addition, parts of the slope are poorly represented in the descending track due to strong atmospheric disturbances, underscoring the challenges of monitoring landslides in mountainous regions near power facilities.

4.2 Time-series deformation characteristics

The InSAR time-series successfully captured both the gradual displacement phase and a period of accelerated movement between April and May 2022, during which significant phase decorrelation occurred due to large deformations (Figures 4a1–a3, b1–b3). Spatially, the average velocity maps derived from ascending and descending tracks (Figures 3a,b) consistently identified an active deformation zone on the western bank of the reservoir, coinciding with the distribution of ground monitoring points (23GP–26GP).

FIGURE 4

Interferograms from both geometries (Figures 4a1–a3, b1–b3) further confirmed the deformation anomalies, where clear fringe discontinuities and distortions were observed in the active slope sector. The descending track data demonstrated greater reliability for this landslide, as it was less affected by temporal decorrelation during the rapid movement period, providing more consistent monitoring results. This reliability is particularly critical for assessing potential threats to nearby infrastructure, such as hydropower facilities and transmission lines.

To evaluate the reliability of the InSAR-derived deformation time series, the results obtained from ascending and descending SAR tracks were compared with independent GNSS monitoring data collected at several benchmark stations within the study area.

The deformation time series derived from the ascending track of Sentinel-1 cover the period from 2 January 2019 to 15 June 2022, while the GNSS observations span from 11 November 2021 to 4 June 2022 for most stations. The exception is station 28GP01, which has a longer monitoring period from 8 June 2021 to 4 June 2022. Figure 5 presents the comparison between the InSAR-derived deformation time series and the GNSS measurements. Overall, the temporal trends of the InSAR results show good agreement with the GNSS observations, indicating that the InSAR time series can reliably capture the deformation evolution of the landslide.

FIGURE 5

Similarly, the deformation time series derived from the descending SAR track cover the period from 4 January 2019 to 5 June 2022, and the corresponding comparison with GNSS measurements is shown in Figure 6. The consistency between the two independent datasets further confirms the robustness of the InSAR-derived deformation signals.

FIGURE 6

It should be noted that the location of GNSS station 26GP01 corresponds to a masked pixel in the InSAR deformation map due to coherence limitations. Therefore, the InSAR time series used for comparison was extracted from the pixel with the shortest Euclidean distance to the GNSS station.

Quantitative accuracy assessment was conducted by calculating the root mean square error (RMSE) between the InSAR-derived deformation and GNSS observations. The RMSE values of the ascending-track results range from 5.01 mm to 13.05 mm, while those of the descending-track results range from 7.07 mm to 29.18 mm. Overall, the ascending-track results show slightly higher consistency with the GNSS observations. These results demonstrate that the InSAR-derived deformation time series provide a reliable representation of ground displacement in the study area.

4.3 Multi-dimensional deformation analysis

By combining ascending and descending Sentinel-1 orbit data under the assumption of a parallel displacement field, two-dimensional deformation rates were successfully estimated. The results revealed significant spatial variability in horizontal movement across the Hongyanzi landslide. As shown in Figure 7a, the mean deformation velocity in the east-west direction reached maximum values of approximately 50 mm/year in the westward direction, particularly concentrated in the central and upper sections of the landslide mass.

FIGURE 7

The north-south deformation component (Figure 7b) exhibited comparatively lower magnitude movements, with the InSAR analysis indicating challenges in precisely constraining this directional movement due to well-known geometrical limitations of the InSAR technique in north-south deformation monitoring. Despite these limitations, the retrieved deformation patterns in both east-west and vertical directions provide valuable quantitative data for assessing potential impacts on critical infrastructure.

The integration of multi-dimensional deformation results demonstrates that the most active movement is occurring in the western direction, coinciding with areas where power transmission towers and access roads are located. This spatial correlation highlights the direct threat posed by continued landslide displacement to the stability and operational safety of regional power infrastructure.

4.4 Large deformation detection using POT

POT analysis using L-band ALOS-2 data provided critical complementary measurements of large-magnitude displacements that exceeded the detection capability of conventional InSAR. As shown in Figure 8, significant offset signals were detected within the Hongyanzi landslide area across multiple interferometric pairs spanning different time intervals between 2018 and 2020.

FIGURE 8

The POT results revealed substantial ground movement patterns that varied temporally and spatially across the landslide body. Figure 8a (20180727_20190531) shows initial displacement signals primarily concentrated in the upper section of the slope. Subsequent pairs (Figure 8b 20180727_20200529; Figure 8c 20181116_20200529) demonstrate the progressive development of deformation, with expanding affected areas and increasing displacement magnitudes. The final pair (Figure 8d 20190531_20200529) captures the cumulative deformation over the longest time span, showing the most extensive displacement pattern.

This approach confirmed the presence of substantial ground movement that could potentially affect the stability of power infrastructure foundations, particularly transmission towers and access roads located in the landslide-affected area. The POT-derived displacements provide valuable information for identifying areas susceptible to rapid sliding events that may pose immediate threats to power system operations.

4.5 Seepage simulation results

Steady-state and transient seepage analyses were conducted to investigate the influence of reservoir water-level fluctuations on the stability of the Hongyanzi landslide. The numerical model, established based on the geological cross-section, consisted of 962 elements and 1058 nodes with refined discretization (5 m for the sliding body and 15 m for the sliding bed) to accurately capture pore water pressure distributions.

As shown in Figure 9a, the model setup followed the geological structure with appropriate boundary conditions. The steady-state simulation (Figure 9b) revealed a predictable groundwater flow pattern consistent with the geological stratification, showing equipotential lines generally parallel to the slope surface with a predominant downward flow direction toward the reservoir.

FIGURE 9

Transient simulations during reservoir drawdown (120 days) and filling (213 days) periods showed significant changes in pore water pressure distribution within the landslide body (Figures 9c,d). During drawdown conditions, the rapid decrease in reservoir level created a hydraulic gradient directed toward the slope exterior, resulting in decreased pore water pressures in the lower portion of the landslide. Conversely, during reservoir filling, the rising water level generated increased pore water pressures in the slope base, potentially reducing the effective stress and shear strength of the sliding zone.

These pore water pressure variations directly influence the stability of the slope and consequently threaten nearby power infrastructure, particularly transmission towers located along the slope crest and access roads crossing the landslide area.

5 Conclusion

Overall, the combined use of satellite-based deformation monitoring and numerical seepage analysis provides complementary information for understanding landslide dynamics in reservoir-affected mountainous environments. While additional observations and validation are required before operational applications can be considered, the proposed framework offers useful insights for future landslide hazard assessment and monitoring strategies in areas influenced by reservoir operations.

References

• 1

  AhmadS.RizviZ.Arsalan KhanM.AhmadJ.WuttkeF. (2019). Experimental study of thermal performance of the backfill material around underground power cable under steady and cyclic thermal loading. Mater. Today Proc.17, 85–95. 10.1016/j.matpr.2019.06.404

  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  AhmadS.RizviZ. H.ArpJ. C. C.WuttkeF.TirthV.IslamS. (2021). Evolution of temperature field around underground power cable for static and cyclic heating. Energies14, 8191. 10.3390/en14238191

  • CrossRef
  • Google Scholar
  • View reference in article

• 3

  AhmadS.RizviZ. H.WuttkeF. (2025a). Unveiling soil thermal behavior under ultra-high voltage power cable operations. Sci. Rep.15, 7315. 10.1038/s41598-025-91831-1

  • CrossRef
  • Google Scholar
  • View reference in article

• 4

  AhmadS.AhmadS.AkhtarS.AhmadF.AnsariM. A. (2025b). Data-driven assessment of corrosion in reinforced concrete structures embedded in clay dominated soils. Sci. Rep.15, 22744. 10.1038/s41598-025-08526-w

  • CrossRef
  • Google Scholar
  • View reference in article

• 5

  AhmadS.RizviZ. H.WuttkeF. (2026). Spatial soil temperature measurement data around an underground power cable under controlled thermal loading. Data Brief65, 112582. 10.1016/j.dib.2026.112582

  • CrossRef
  • Google Scholar
  • View reference in article

• 6

  AswathiJ.Binoj KumarR. B.OommenT.BoualiE. H.SajinkumarK. S. (2022). InSAR as a tool for monitoring hydropower projects: a review. Energy Geosci.3, 160–171. 10.1016/j.engeos.2021.12.007

  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  BatesonL.CignaF.BoonD.SowterA. (2015). The application of the intermittent SBAS (ISBAS) InSAR method to the south Wales coalfield, UK. Int. J. Appl. Earth Observation Geoinformation34, 249–257. 10.1016/j.jag.2014.08.018

  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  BerardinoP.FornaroG.LanariR.SansostiE. (2002). A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms. IEEE Trans. Geosci. Remote Sens.40, 2375–2383. 10.1109/TGRS.2002.803792

  • CrossRef
  • Google Scholar
  • View reference in article

• 9

  BertoliniG.GuidaM.PizzioloM. (2005). Landslides in Emilia-Romagna region (Italy): strategies for hazard assessment and risk management. Landslides2, 302–312. 10.1007/s10346-005-0020-1

  • CrossRef
  • Google Scholar
  • View reference in article

• 10

  CignaF.Esquivel RamírezR.TapeteD. (2021). Accuracy of Sentinel-1 PSI and SBAS InSAR displacement velocities against GNSS and geodetic leveling monitoring data. Remote Sens.13, 4800. 10.3390/rs13234800

  • CrossRef
  • Google Scholar
  • View reference in article

• 11

  DaiF. C.LeeC. F.NgaiY. Y. (2002). Landslide risk assessment and management: an overview. Eng. Geol.64, 65–87. 10.1016/S0013-7952(01)00093-X

  • CrossRef
  • Google Scholar
  • View reference in article

• 12

  ElmoreA.StylinskiC.PradhanK. (2016). Synergistic use of citizen science and remote sensing for continental-scale measurements of forest tree phenology. Remote Sens.8, 502. 10.3390/rs8060502

  • CrossRef
  • Google Scholar
  • View reference in article

• 13

  FanZ. H.WangS. M.WangL.GuoF.NanF.HuY. (2024). Emergency treatment effect evaluation of rear-slope cutting and front-slope pressing on a hydrodynamic pressure landslide: a case study of the shuping landslide in the three gorges reservoir area. Bull. Eng. Geol. Environ.83 (1), 38. 10.1007/s10064-023-03539-z

  • CrossRef
  • Google Scholar
  • View reference in article

• 14

  FerrettiA.PratiC.RoccaF. (2001). Permanent scatterers in SAR interferometry. IEEE Trans. Geosci. Remote Sens.39, 8–20. 10.1109/36.898661

  • CrossRef
  • Google Scholar
  • View reference in article

• 15

  GodtJ. W.Şener‐KayaB.LuN.BaumR. L. (2012). Stability of infinite slopes under transient partially saturated seepage conditions. Water Resour. Res.48, 2011WR011408. 10.1029/2011WR011408

  • CrossRef
  • Google Scholar
  • View reference in article

• 16

  HanB.TongB.YanJ.YinC.ChenL.LiD. (2018). The monitoring-based analysis on deformation-controlling factors and slope stability of reservoir landslide: hongyanzi landslide in the southwest of China. Geofluids2018, 1–14. 10.1155/2018/7391517

  • CrossRef
  • Google Scholar
  • View reference in article

• 17

  HaroonA.AhmadS.HussainA. (2017). “CFD prediction of loss coefficient in straight pipes,” in Development of water resources in India. Water science and technology library. Editors GargV.SinghV. P.RajV. (Cham: Springer International Publishing), 75, 477–485. 10.1007/978-3-319-55125-8_41

  • CrossRef
  • Google Scholar
  • View reference in article

• 18

  HermanR. L.RobinsonW. A.FrankeS. J. (1999). Observational evidence of quasi two‐Day/gravity wave interaction using MF radar. Geophys. Res. Lett.26, 1141–1144. 10.1029/1999GL900157

  • CrossRef
  • Google Scholar

• 19

  HuangH.LiuC.WangX.ZhouX.GongP. (2019). Integration of multi-resource remotely sensed data and allometric models for forest aboveground biomass estimation in China. Remote Sens. Environ.221, 225–234. 10.1016/j.rse.2018.11.017

  • CrossRef
  • Google Scholar

• 20

  HuangC.YinK.LiangX.GuiL.ZhaoB.LiuY. (2024). Study of direct and indirect risk assessment of landslide impacts on ultrahigh-voltage electricity transmission lines. Sci. Rep.14, 25719. 10.1038/s41598-024-76726-x

  • CrossRef
  • Google Scholar

• 21

  JinR.LiX.CheT. (2009). A decision tree algorithm for surface soil freeze/thaw classification over China using SSM/I brightness temperature. Remote Sens. Environ.113, 2651–2660. 10.1016/j.rse.2009.08.003

  • CrossRef
  • Google Scholar

• 22

  LauR.SeguíC.WatermanT.ChaneyN.VeveakisM. (2024). InSAR-Informed in situ monitoring for deep-seated landslides: insights from El forn (Andorra). Nat. Hazards Earth Syst. Sci.24, 3651–3661. 10.5194/nhess-24-3651-2024

  • CrossRef
  • Google Scholar

• 23

  LiuG. S.WangB.SunQ.HuJ.LiuL. L.ZhengW. J.et al (2025). New insights into the reservoir landslide deformation mechanism from InSAR and numerical simulation technology. IEEE J. Sel. Top. Appl. Earth Observations Remote Sens.18, 2908–2927. 10.1109/jstars.2024.3523294

  • CrossRef
  • Google Scholar

• 24

  LiuG. S.SunQ.HuJ.LiuL.-L.ZhengW. J.HanB.et al (2026). A novel step-like deformation model for reservoir landslide monitoring with multi-temporal InSAR. Sci. China Earth Sci.69 (2), 679–701. 10.1007/s11430-025-1780-2

  • CrossRef
  • Google Scholar

• 25

  Monti-GuarnieriA.ManzoniM.GiudiciD.RecchiaA.TebaldiniS. (2020). Vegetated target decorrelation in SAR and interferometry: models, simulation, and performance evaluation. Remote Sens.12, 2545. 10.3390/rs12162545

  • CrossRef
  • Google Scholar

• 26

  PanB.ShiX. (2023). Fusing ascending and descending time-series SAR images with dual-polarized pixel attention UNet for landslide recognition. Remote Sens.15, 5619. 10.3390/rs15235619

  • CrossRef
  • Google Scholar

• 27

  PelasciniL.SteerP.MouyenM.LonguevergneL. (2022). Finite-hillslope analysis of landslides triggered by excess pore water pressure: the roles of atmospheric pressure and rainfall infiltration during typhoons. Nat. Hazards Earth Syst. Sci.22, 3125–3141. 10.5194/nhess-22-3125-2022

  • CrossRef
  • Google Scholar

• 28

  PetleyD. (2012). Global patterns of loss of life from landslides. Geology40, 927–930. 10.1130/G33217.1

  • CrossRef
  • Google Scholar

• 29

  RizviZ. H.MustafaS. H.SattariA. S.AhmadS.FurtnerP.WuttkeF. (2020). “Dynamic lattice element modelling of cemented geomaterials,” in Advances in computer methods and geomechanics. Lecture notes in civil engineering. Editors PrashantA.SachanA.DesaiC. S. (Singapore: Springer Singapore), 55, 655–665. 10.1007/978-981-15-0886-8_53

  • CrossRef
  • Google Scholar

• 30

  SolariL.Del SoldatoM.RaspiniF.BarraA.BianchiniS.ConfuortoP.et al (2020). Review of satellite interferometry for landslide detection in Italy. Remote Sens.12, 1351. 10.3390/rs12081351

  • CrossRef
  • Google Scholar

• 31

  SunD.DengW.YangT.LiJ.ZhaoY. (2023). A case study integrating numerical simulation and InSAR monitoring to analyze bedding-controlled landslide in nanfen open-pit mine. Sustainability15, 11158. 10.3390/su151411158

  • CrossRef
  • Google Scholar

• 32

  TangF.QiS.GuoS.LiY.GuoX.LuX.et al (2025). The influence of reservoirs on landslide erosion. Remote Sens.17, 569. 10.3390/rs17040569

  • CrossRef
  • Google Scholar

• 33

  TorresR.SnoeijP.GeudtnerD.BibbyD.DavidsonM.AttemaE.et al (2012). GMES Sentinel-1 mission. Remote Sens. Environ.120, 9–24. 10.1016/j.rse.2011.05.028

  • CrossRef
  • Google Scholar

• 34

  TresoldiG.ArosioD.HojatA.LongoniL.PapiniM.ZanziL. (2019). Long-term hydrogeophysical monitoring of the internal conditions of river levees. Eng. Geol.259, 105139. 10.1016/j.enggeo.2019.05.016

  • CrossRef
  • Google Scholar

• 35

  VakaD. S.YaragundaV. R.PerdikouS.PapanicolaouA. (2024). InSAR integrated machine learning approach for landslide susceptibility mapping in California. Remote Sens.16, 3574. 10.3390/rs16193574

  • CrossRef
  • Google Scholar

• 36

  ValadeS.DonnadieuF.LesageP.MoraM. M.HarrisA.AlvaradoG. E. (2012). Explosion mechanisms at arenal volcano, Costa Rica: an interpretation from integration of seismic and doppler radar data. J. Geophys. Res.117, 2011JB008623. 10.1029/2011JB008623

  • CrossRef
  • Google Scholar

• 37

  WrightT. J.ParsonsB. E.LuZ. (2004). Toward mapping surface deformation in three dimensions using InSAR. Geophys. Res. Lett.31, 2003GL018827. 10.1029/2003GL018827

  • CrossRef
  • Google Scholar

• 38

  XuS.ZhouX. (2025). Spatiotemporal evolution characteristics of hanyuan landslide in Sichuan Province, China, on 21 August 2020. Appl. Sci.15, 3872. 10.3390/app15073872

  • CrossRef
  • Google Scholar

• 39

  YuP.ShiW.CaoZ.CaoX.WangR.WuW.et al (2024). Numerical analysis of seepage field response characteristics of weathered granite landslides under fluctuating rainfall conditions. Water16, 1996. 10.3390/w16141996

  • CrossRef
  • Google Scholar

• 40

  YueX.WangY.WenT. (2024). An inversion study of reservoir colluvial landslide permeability coefficient by combining physical model and data-driven models. Water16, 686. 10.3390/w16050686

  • CrossRef
  • Google Scholar

• 41

  ZouZ.TangH.CrissR. E.HuX.XiongC.WuQ.et al (2021). A model for interpreting the deformation mechanism of reservoir landslides in the three gorges reservoir area, China. Nat. Hazards Earth Syst. Sci.21, 517–532. 10.5194/nhess-21-517-2021

  • CrossRef
  • Google Scholar