Ground deformation monitoring of landslides and precipitation-triggered mechanisms using GBInSAR

Landslides are common in mountainous, hilly, and steep-slope regions, posing serious threats to human life, infrastructure, and ecological systems. Precipitation is a major triggering factor that can rapidly destabilize slopes by increasing moisture content and reducing soil shear strength. This study proposes an integrated landslide monitoring framework that combines satellite-borne InSAR for regional-scale detection, GB-InSAR for high-density dynamic monitoring, GNSS for precise three-dimensional ground deformation constraints, and rain-gauge observations for precipitation forcing characterization, with the aim of improving monitoring accuracy, spatiotemporal resolution, and early-warning capability. The results reveal a clear temporal correspondence between rainfall events and landslide displacement: during precipitation peaks (0.4 mm and 0.3 mm), displacement increased to approximately +25 mm and +35 mm, indicating direct rainfall-driven deformation responses. GB-InSAR further captures pronounced spatial heterogeneity across the landslide body, with maximum cumulative displacement reaching +656.3 mm and minimum displacement −1098.17 mm. GNSS measurements corroborate these deformation patterns, particularly during the acceleration stage when the trends derived from GNSS and GB-InSAR are highly consistent. Overall, multi-source evidence indicates that both short-duration intense rainfall and long-term precipitation accumulation contribute to landslide activity, with precipitation capable of inducing not only rapid displacement surges but also slower deformation under low-rainfall conditions. These findings provide a robust basis for landslide early warning and risk assessment in regions frequently affected by intense rainfall and sustained wet periods.

1 Introduction

Landslide disasters are one of the most common geological hazards globally, particularly in areas prone to such events, such as mountainous, hilly, and slope regions (Yang et al., 2020). Landslides pose significant threats to human life, property safety, and the ecological environment. Precipitation (Abgrami et al., 2025), as one of the key triggering factors of landslide disasters, can lead to slope instability in a short period by influencing the moisture content of the slope and weakening the soil’s shear strength. Studying the relationship between precipitation and landslide deformation, and exploring the mechanisms by which precipitation triggers landslides, is crucial for early warning and prevention of landslide disasters (Su et al., 2022).

Traditional landslide monitoring methods primarily rely on ground surveys, drilling tests, and optical remote sensing imagery. While these approaches provide useful information (Qiu et al., 2020), they are often constrained by limited spatial coverage, discontinuous observations, and insufficient real-time capability, especially for capturing early-stage deformation and rapid kinematic changes (Hamedi et al., 2025). With advances in SAR remote sensing, differential interferometric synthetic aperture radar (DInSAR) and ground-based InSAR (GB-InSAR) have become key tools for deformation monitoring. GB-InSAR, for example (Ferrigno et al., 2017), enables high-resolution deformation measurements by acquiring detailed radar images (Lingua et al., 2008), and can maintain monitoring performance under complex meteorological conditions, making it highly promising for landslide surveillance (Tian et al., 2019).

However, despite the strong capability of GB-InSAR for high-precision deformation monitoring, achieving comprehensive landslide monitoring through effective integration with complementary techniques remains challenging. To address this gap, this study combines satellite-borne InSAR surveys with ground-based GB-InSAR observations: satellite-borne InSAR is used for large-scale landslide screening and hazard identification, and GB-InSAR is then deployed in the selected hotspots for high-density, dynamic deformation tracking. In addition, GNSS (Global Navigation Satellite System) comparative monitoring is incorporated to provide more reliable ground deformation constraints, and rainfall gauge data are integrated to examine precipitation-driven triggering processes by linking rainfall forcing with deformation evolution and the associated hydrological and strength-weakening mechanisms (Woods et al., 2021). The framework is demonstrated in the Longmenshan region, Sichuan Province, China, located at the transition between the western Sichuan Basin margin and the Longmenshan Mountains (Ebadati et al., 2025), where steep terrain and frequent rainfall-related landslides and debris flows are well documented, particularly during heavy rainfall periods (Moretto et al., 2021). In this region, precipitation is widely considered a primary trigger, especially under continuous rainfall or short-duration intense storms (Birjandi and Derakhshani, 2025) NIDE.

Unlike existing research in landslide monitoring, this study addresses the limitations of traditional monitoring methods in spatial and temporal resolution by integrating multiple advanced technologies (Ahadi and Rosta, 2024). First, satellite-borne InSAR is used for large-scale surveys of landslide areas, quickly identifying potential landslide hazard zones and providing data support for subsequent high-precision on-site monitoring. Second, GB-InSAR equipment is deployed in landslide-prone areas and combined with GNSS for three-dimensional line-of-sight (LOS) comparative monitoring, enabling real-time tracking of dynamic changes in landslides and capturing the temporal and spatial relationship between precipitation events and landslide deformations. Finally, by integrating rainfall gauge data, the study analyzes the role of precipitation in landslide occurrences and elucidates the mechanism by which precipitation triggers deformation by altering slope hydrological characteristics and weakening soil strength (Schlögel et al., 2015; Yin et al., 2010; Zhang et al., 2020; Yeganeh et al., 2025).

In summary, a major limitation in precipitation-driven landslide studies is the difficulty of simultaneously achieving broad spatial coverage and continuous, high-density temporal monitoring. Satellite-borne InSAR is effective for regional screening but remains constrained by revisit intervals and viewing geometry, whereas GB-InSAR provides near-real-time, high-resolution deformation measurements but is typically limited to site-scale deployment; consequently, single-sensor approaches often fail to robustly characterize event-scale rainfall–deformation responses, including potential lag effects and stage-wise evolution. To address this gap, this study develops a space–ground integrated monitoring workflow that links regional satellite InSAR screening with targeted GB-InSAR deployment, and further couples GB-InSAR displacement time series with independent GNSS constraints and rain-gauge observations to form a temporally consistent, event-oriented interpretation of rainfall-triggered deformation. By explicitly using GNSS as an external check on GB-InSAR observations and relating deformation accelerations to rainfall forcing under real monitoring conditions, the proposed framework provides a practical pathway to improve monitoring reliability and strengthen the early-warning relevance of precipitation-driven landslide assessment.

2 Background and features of landslide

The study area is located in the Longmenshan region of Sichuan Province, China, situated at the transition zone between the western edge of the Sichuan Basin and the Longmenshan Mountains, exhibiting typical mountainous geological features. This region is renowned for frequent landslides and debris flow disasters, especially during periods of heavy rainfall, when landslides and debris flows are particularly prominent. The Longmenshan area has a complex topography with steep slopes and loose soil, and the high precipitation levels make the slope prone to sliding and collapsing under the influence of water, resulting in severe natural disasters.

As shown in Figure 1, the study area lies to the west of Chengdu, Sichuan Province, at an elevation of approximately 2,000 m, within the middle and lower mountain ranges of the Longmenshan Mountains. The terrain is highly undulating, primarily consisting of hills and mountains. The rock formations of the mountains are complex, with loose rock and soil, making the area susceptible to natural factors that trigger landslides, debris flows, and other disasters. The geological background and topographical features of the study area make it highly prone to ground deformation, particularly due to the frequent occurrence of debris flows.

Figure 1

Figure 1. Location of the study area.

The Longmenshan region is heavily influenced by strong precipitation throughout the year, particularly during the rainy season (usually in summer and autumn). The large amounts of rainfall often lead to the saturation of the slope’s moisture content, triggering debris flows and landslides. Debris flows are especially severe in this region and are one of the main types of geological disasters. These flows predominantly occur on steep slopes, in valleys with significant gradients, and in the upper reaches of drainage basins. After heavy rainfall, water accumulation on the mountains rapidly infiltrates the soil layers, causing a sharp decrease in soil strength, which leads to the formation and movement of debris flows. These flows not only threaten the safety of local residents’ lives and property but also cause significant damage to transportation, infrastructure, and other assets. Historical records show that the region has experienced several severe debris flow disasters in recent years. For example, in 2011 and 2013, consecutive heavy rainfall events triggered large-scale debris flows, resulting in extensive damage to houses and roads and causing numerous casualties. The occurrence of debris flows is not only directly related to the amount of rainfall but also influenced by various factors such as regional slope, soil type, and vegetation coverage.

Although landslides and debris flows are common natural disasters in this region, traditional monitoring methods often fail to capture the dynamic changes of these disasters in a timely manner. Conventional methods, such as manual surveys, localized monitoring, and geological exploration, are typically limited to small-scale or short-term data collection, making it difficult to conduct long-term, real-time monitoring over large areas. Therefore, the research and application of new remote sensing technologies, especially differential interferometry synthetic aperture radar (DInSAR) and the improved GB-InSAR technology, are of great significance for improving the accuracy of landslide and debris flow monitoring and achieving disaster early warning.

Figure 2 shows the locations of various monitoring devices deployed in the study area. It illustrates the distribution of GB-InSAR radar monitoring regions, GNSS stations and sites, rain gauges, and other equipment, providing multi-level and multi-dimensional data support for landslide and debris flow monitoring in this study. The radar monitoring area, located in the upper-right section of the figure, covers the primary landslide and debris flow monitoring regions within the study area. The radar system in this region, through high-resolution ground deformation monitoring, can capture real-time ground displacement changes before and after landslides, offering critical data for analyzing the mechanisms behind precipitation-triggered landslide disasters. Furthermore, the selection of the radar monitoring area is based on historical landslide data and topographical features, ensuring that the monitoring region covers areas with high landslide risk.

Figure 2

Figure 2. Equipment location deployment diagram.

In the center of the figure, GNSS base stations are placed at the core location of the study area. Data transmission between this base station and several GNSS sites (GNSS Station 1, GNSS Station 2) allows for real-time collection of ground displacement information, assisting in the analysis of landslide dynamics. The layout of GNSS sites primarily focuses on high-risk landslide and debris flow areas, and the high-precision positioning data obtained are compared with GB-InSAR monitoring results, providing accurate references for the quantitative analysis of ground deformation. Additionally, rain gauges (Rain Gauge 1, Rain Gauge 2, Rain Gauge 3, and Rain Gauge 4) are installed at several key locations within the study area. These rain gauges are used to monitor precipitation in real time, especially in relation to the timing of landslide and debris flow events, revealing the temporal and spatial relationship between precipitation and ground deformation. Through the collection of precipitation data, the study can further explore how precipitation influences soil moisture, slope stability, and the occurrence of landslides.

The landslide-prone areas indicated on the figure highlight regions where landslide disasters are likely to occur. The monitoring data from these regions are critical for this study. By integrating data from GB-InSAR, GNSS, and rain gauges, the study enables comprehensive monitoring of landslide and debris flow occurrence and evolution, providing reliable technological means for landslide disaster early warning.

3 Methodology

This study primarily utilizes SBAS-InSAR technology, a time-series InSAR method for processing multi-band satellite data, first proposed by the Berardino team in 2002 (Shi et al., 2021; Wentao et al., 2021; Yao et al., 2022). This technology, through the use of multiple master images and singular value decomposition (SVD), effectively captures the time-series deformation rates within the study area. Compared to traditional D-InSAR technology, this method demonstrates excellent performance in reducing spatial-temporal decorrelation. The data processing includes several stages: first, the raw data is converted into SLC (Single Look Complex) format, followed by the setting of time and spatial baseline thresholds. Based on the connectivity relationships, interferograms are generated for each interferometric pair, followed by flattening, adaptive filtering, and phase unwrapping (Cai et al., 2022; Gorsevski and Jankowski, 2010; Lu and Zeng, 2020). Next, ground control point data is used to remove residual topographic phases, and further phase unwrapping and atmospheric phase correction are performed. Finally, through the geo-referencing process, deformation time-series data in a geographic coordinate system is obtained (Del Soldato et al., 2021).

For ground-based InSAR monitoring, this study uses the R/HYB2000 MIMO slope radar system, produced by Suzhou Leko Sensor Technology Co., Ltd. The instrument, developed by Beijing Institute of Technology, operates in the Ku-band with a central frequency of 17.2 GHz. It offers high spatial resolution and strong anti-interference capabilities, making it suitable for landslide monitoring in complex terrains (Karunathilake et al., 2019). The radar system’s spatial resolution is 0.5 m × 0.5 m, enabling precise detection of minute deformations in the landslide mass (Mugnai and Tarchi, 2022). For the parameters of the employed radar system, see Table 1. It also provides rapid temporal resolution, acquiring four full-field interferograms per hour, with a swath width of 1.5 km × 1.0 km. The monitoring period is set from January to June 2025, with the radar system installed on stable bedrock on the opposite side of the landslide, approximately 400 m away from the landslide mass (Liu et al., 2021), effectively avoiding the impact of the landslide’s instability on the radar system. The equipment operates based on MIMO (Multiple Input Multiple Output) technology, allowing multiple antenna arrays to simultaneously transmit and receive radar waves, thus improving data collection efficiency and enhancing the ability to process complex signals. After each scan, atmospheric phase screen (APS) compensation and phase unwrapping techniques are applied to improve deformation monitoring accuracy. Moreover, the system uses dynamic deformation extraction algorithms to monitor real-time deformation changes in the landslide mass, providing high temporal resolution, particularly for detecting rapid landslide changes and identifying acceleration characteristics.

Table 1

Table 1. Parameters of the employed radar system.

In addition, this study adopts a combined approach using GNSS and GB-InSAR monitoring data to achieve precise analysis of ground deformation in the landslide area (Acar et al., 2008; Liao et al., 2022; Senogles et al., 2023). Since GNSS measurements provide three-dimensional deformation data in the eastward, northward, and vertical directions, while the radar system measures radial deformation, it is necessary to convert the three-dimensional deformation data obtained from GNSS into the radial deformation components along the radar line of sight. The converted radial deformation components can then be compared with the radar monitoring data to further reveal the dynamic processes of the landslide disaster. The following outlines the specific analysis method.

Figures 3a–c illustrate the schematic diagrams for considering only the northward/eastward/vertical deformations. In these figures, the left image shows a plan view, while the right image provides a three-dimensional representation of the deformation. Point ‘a’ represents the radar, point ‘a’ is the projection of the radar on the monitoring point’s plane, point ‘b’ is the position of the monitoring point, and point c is the position of the monitoring point after deformation. x1 is the deformation in the northward direction as measured by GNSS, r is the projection of the radar line of sight on the plane, θ is the angle between the northward deformation direction at the monitoring point and the radar’s radial direction, and α is the angle between the deformation direction (north) at the monitoring point and the radar’s normal.

Figure 3

Figure 3. GNSS East-North-Up direction conversion to GBINSAR LOS direction illustration. (a) illustrates the schematic diagram of transforming the GNSS north-component displacement into the radar line-of-sight (LOS) direction; (b) illustrates the schematic diagram of transforming the GNSS east-component displacement into the radar LOS direction; and (c) illustrates the schematic diagram of transforming the GNSS vertical-component displacement into the radar LOS direction.

In triangle Δbcd, applying the cosine rule of the triangle it is known

$\cos \mathrm{\theta }=\frac{{\text{bd}}^2+{\text{bc}}^2-{\text{cd}}^2}{2\text{bd}·\text{bc}}(1)$

In Equation 1, $\text{bc}=x_1$, $\text{bd}=\frac{r·x_1\mathit{cos}\alpha }{r^{\prime }}$, $\text{cd}=\sqrt{{\left(x_1\mathit{sin}\alpha \right)}^2+{\left(\frac{\mathrm{h}·x_1\mathit{cos}\alpha }{r^{\prime }}\right)}^2}$.

The northward deformation component in the radar line of sight direction is

${\Delta \mathrm{x}}_1=\frac{{{\mathrm{x}}_1}^2+{\left(\frac{\left(\mathrm{r}{\mathrm{x}}_1\cos \mathrm{\alpha }\right)}{{\mathrm{r}}^{\prime }}\right)}^2-{\left({\mathrm{x}}_1\sin \mathrm{\alpha }\right)}^2-{\left(\frac{\mathrm{h}{\mathrm{x}}_1\cos \mathrm{\alpha }}{{\mathrm{r}}^{\prime }}\right)}^2}{2\frac{\mathrm{r}{\mathrm{x}}_1\cos \mathrm{\alpha }}{{\mathrm{r}}^{\prime }}}={\mathit{x}}_1\mathit{cos}\mathrm{\alpha }·\frac{{\mathit{r}}^{\prime }}{\mathit{r}}(2)$

In Equation 2, $r^{\prime }=\sqrt{r^2-{\mathrm{h}}^2}$.

The deformation component of the GNSS-measured deformation in the radar-monitoring point radial direction

$\Delta \mathrm{x}={\Delta \mathrm{x}}_1+{\Delta \mathrm{x}}_2+{\Delta \mathrm{x}}_3=\frac{{\mathrm{r}}^{\prime }\left({\mathrm{x}}_1\cos \mathrm{\alpha }+{\mathrm{x}}_2\sin \mathrm{\alpha }\right)+\mathrm{h}{\mathrm{x}}_3}{\mathrm{r}}(3)$

In Equation 3, Δx2 is the eastward deformation component from GNSS, and Δx3 is the vertical deformation component from GNSS.

4 Result

4.1 Satellite-borne InSAR long-term time-series landslide displacement monitoring results

Figure 4 presents the long-term time-series displacement monitoring results of the landslide area from January 2024 to May 2025, using the SBAS (Small Baseline Subset) method. The subplots from (a) to (f) show the cumulative displacement changes in the region over different time periods.

Figure 4

Figure 4. The SBAS InSAR monitoring results: (a–f) represent the accumulated deformation results for the monitoring period from January 2024 to May 2025.

In the figure, the red areas represent regions moving away from the satellite (subsidence), while the blue areas represent regions moving towards the satellite (uplift). Specifically, in Figure 4a, the green areas indicate zones with little or no significant deformation, typically representing geologically stable areas with minimal displacement. Over time, from Figures 4b–f, the displacement range gradually increases. After December 2024, signs of collapse begin to appear in the lower-left section, with some specific areas showing displacements reaching several tens of millimeters, indicating that landslide activity in these regions is gradually intensifying. The colors represent displacement, with the unit being millimeters (mm). The time periods for each subplot are indicated in the legend within the figure, such as Figures 4c,d, where 20240922 refers to 22 September 2024.

The geological mechanisms underlying these changes are generally closely related to factors such as precipitation, geological structure, and soil conditions. Heavy rainfall can cause groundwater levels to rise or soil to become saturated, activating sliding surfaces within the landslide mass and leading to more significant ground deformation. The progressively increasing displacement changes shown in the figure, especially the displacements along the landslide centerline, are likely to be closely associated with the influence of these natural factors.

4.2 GBInSAR ground displacement monitoring results

Figure 5 presents the ground displacement monitoring results for the landslide area using GBInSAR technology. This technique, through high-resolution time-series interferometry, provides precise displacement data, capturing millimeter-scale ground deformations, making it especially suitable for monitoring landslides and other disaster activities. The GB-InSAR measures one-dimensional displacement along the radar line-of-sight (LOS), i.e., the projection of the true ground motion onto the viewing direction of the instrument. The system used in this study is a ground-based MIMO (multiple-input multiple-output) radar, which forms the interferometric phase by transmitting/receiving with multiple antenna channels and repeatedly imaging the same slope scene; consequently, the derived displacement represents LOS-projected motion rather than full 3-D movement. The displacement data shown in the figure reveal significant spatial heterogeneity across the region, reflecting the deformation differences of various geological units in both time and space. Within the monitoring area, there is a considerable variation in displacement amplitude, and through analysis of the monitoring results, the area can be divided into two main sliding zones.

Figure 5

Figure 5. GB-InSAR line-of-sight (LOS) displacement map of the monitored landslide. Positive values indicate motion toward the radar (range shortening), and negative values indicate motion away from the radar (range lengthening).

The first sliding zone is located in the catchment and river channel areas, where significant negative displacement is observed, particularly after precipitation, when the displacement increases notably. The accumulation of water triggered by rainfall, soil saturation, and rising groundwater levels may be the primary factors contributing to the intensified landslide activity in these areas. The wetting effect of precipitation on the soil reduces its strength, leading to displacement changes in the landslide mass. Given the instrument viewing geometry, the prevalent negative LOS displacement in this zone suggests that the dominant motion component is projected away from the radar, which is consistent with downslope movement whose vector has an outward LOS component for the monitored slope sector; conversely, localized positive LOS displacement indicates motion components projected toward the radar, reflecting either different slope-facing directions or locally varying kinematics within the channel deposits. The displacement amplitude in the river channel area is relatively large, showing a clear alternation of positive and negative displacements. This may be related to the erosional effects of river flow and fluctuations in groundwater, which further exacerbate the dynamic changes of the landslide. The significant displacement changes indicate that hydrological conditions play a dominant role in this sliding zone.

The second sliding zone is primarily distributed in the terrace areas. Although the displacement in this region is smaller and the overall geological characteristics appear stable, some localized areas still show slight displacement changes. The relative stability of the terrace area may be attributed to the presence of hard rock layers and less hydrological influence. However, the occurrence of local displacements suggests that groundwater infiltration or other geological activities may still play a role in this region. Although the terrace area shows relatively small ground deformation overall, attention should still be paid to the potential landslide risk, especially during periods of concentrated rainfall.

In summary, the displacement data from the monitoring area reveal two main sliding zones, located in the catchment area, river channel, and terrace areas. The displacement changes in these two sliding zones are closely related to the geological structure and topographical conditions, as well as hydrological factors, particularly precipitation and groundwater. Through high-precision monitoring with GBInSAR technology, critical data support can be provided for early warning, risk assessment, and the development of mitigation measures for landslide disasters.

5 Discussion

5.1 Triggering effect of precipitation events on landslides

To further analyze the relationship between precipitation events and landslide displacement, Figure 6 compares precipitation monitoring results with GBInSAR monitoring data. The average values from four rain gauges over the same time period are selected. It is clearly evident from the figure that there is a significant temporal correlation between precipitation and landslide displacement. The rainfall refers to the 10-min rainfall amount (mm per 10 min), averaged from four gauges, while the orange curve indicates displacement in the landslide area (in mm). By comparing these two datasets, we can clearly observe the triggering effect of precipitation events on landslide displacement, particularly during periods of heavy rainfall, when displacement changes are more pronounced.

Figure 6

Figure 6. Comparison between rain gauge monitoring results and GBInSAR monitoring results.

It should be noted that Figure 6 is presented as a representative example of a typical precipitation-induced deformation episode, rather than a summary of the full monitoring window. The long-term deformation evolution of the landslide from January 2024 to May 2025 is provided in Figure 4, whereas Figure 6 focuses on a short, well-defined rainfall sequence in February 2025 to illustrate the event-scale coupling between rainfall forcing and deformation response. During this episode, two rainfall peaks occurred on 6 February 2025 and 10 February 2025. When precipitation rapidly increased to 0.4 mm on February 6, the landslide displacement rose sharply to approximately +25 mm, suggesting an immediate deformation response to intense short-duration rainfall. A similar behavior was observed on 10 February 2025, when precipitation reached ∼0.3 mm and displacement increased to ∼+35 mm, further supporting a strong rainfall–deformation linkage during concentrated precipitation. Note that the reported 25–35 mm refers to the cumulative GB-InSAR displacement along the radar line-of-sight (LOS) over the event window, rather than an instantaneous step at a single epoch.

The monitored slope is located at the outlet of an old debris-flow gully, where loose deposits and preferential flow pathways can facilitate rapid infiltration and localized pore-pressure rise. Under antecedent rainfall accumulation and/or a near-saturated slope condition, short-duration rainfall bursts can further reduce effective stress and shear strength, leading to accelerated cumulative motion. Beyond the temporal correspondence, the spatial pattern of LOS displacement in Figure 5 suggests a clear kinematic partitioning of the landslide body into hydrologically and structurally distinct domains, including the catchment, channel, and terrace sectors. The catchment zone tends to behave as an upslope recharge and water-convergence area, where runoff concentration and subsurface flow convergence can promote rapid wetting and transient pore-pressure increases during rainfall episodes. In contrast, the channel zone is dominated by gully-confined deposits and repeated remobilization of loose materials, making it more sensitive to short-lived hydrological perturbations and local toe conditions. The terrace sector, characterized by comparatively gentler morphology and likely stronger or more intact material, generally shows a more buffered response, with deformation expressed as localized patches rather than a coherent, slope-wide acceleration. However, during periods of lower precipitation, such as 2 February 2025, and 4 February 2025, landslide displacement changes were small, typically between −5 mm and +5 mm, indicating that landslide activity remained stable and was not significantly triggered by precipitation. During these times, precipitation levels were generally below 0.2 mm, and the soil wetting effect was relatively slow, failing to trigger noticeable acceleration in displacement. This phenomenon suggests that while precipitation significantly influences landslides, its effect needs to reach a certain intensity and duration to trigger substantial displacement responses.

The blue box regions marked in the figure highlight periods of concentrated precipitation, where the acceleration of landslide displacement is closely related to precipitation peaks. Particularly after a sharp increase in precipitation, displacement changes are most pronounced. At the same time, the “Slow deformation” region marked in the figure shows that even under lower precipitation conditions, the landslide mass still experienced slow displacement changes. Even during periods of minimal rainfall, the continuous wetting effect of the soil and gradual rise in groundwater levels led to chronic deformation of the landslide mass, indicating that long-term moisture accumulation has a sustained impact on landslide activity.

By integrating GBInSAR technology with precipitation monitoring data, we can more precisely identify the spatiotemporal evolution patterns of landslide displacement triggered by precipitation events. Precipitation not only triggers rapid landslide displacement but also causes slow displacement changes during lower precipitation periods. The combined effects of short-duration intense rainfall and long-term accumulated precipitation serve as significant inducements for landslide activity. This finding provides a more scientific and quantitative basis for early warning and mitigation of landslide disasters, especially in the context of frequent extreme weather events. Monitoring the relationship between precipitation and landslide displacement is therefore of critical importance.

5.2 Accuracy validation of GBInSAR and GNSS in landslide monitoring

In addition, this study adopts a combined approach using GNSS and GB-InSAR monitoring data to achieve precise analysis of ground deformation in the landslide area. Since GNSS measurements provide three-dimensional deformation data in the eastward, northward, and vertical directions, while the radar system measures line-of-sight deformation, it is necessary to convert the three-dimensional deformation data from GNSS into the deformation components along the radar line of sight. The converted line-of-sight deformation components can then be compared with the radar monitoring data, further revealing the dynamic processes of the landslide disaster.

Figure 7 illustrates the comparison of acceleration changes from GBInSAR and GNSS monitoring data, further validating the consistency of the two monitoring methods in landslide monitoring. The pink curve represents the acceleration measured by GNSS (in mm/h²), while the orange curve represents the acceleration from GBInSAR (in mm/h²). From the figure, it is evident that the acceleration trends from GNSS and GBInSAR are generally consistent, especially during stages where the landslide experiences deformation at different rates. The acceleration changes of both methods show similar trends in the “chronic deformation” and “accelerated deformation” phases, indicating that both monitoring methods can reflect the dynamic response of the landslide mass.

Figure 7

Figure 7. GBInSAR and GNSS deformation verification).

However, despite the consistent trends, there are some differences in the specific magnitudes of acceleration. The primary cause of this difference is related to the working principles of the two monitoring methods. The GNSS system receives ground motion data via satellite signals and typically has high spatial resolution, but it is more susceptible to atmospheric delays, signal blockage, and other factors. Particularly in complex terrain, the accuracy of GNSS may decrease. In contrast, GBInSAR technology is based on the interferometric principle of radar beams, obtaining displacement data by receiving radar signals reflected from the ground. It allows for high-precision monitoring over large areas, especially in large-scale landslide regions, where it holds distinct advantages. However, the line-of-sight measurement characteristic of the radar system means that the displacement measured by GBInSAR is typically related to the line of sight, which may lead to differences in the magnitude of acceleration compared to GNSS, especially when there are changes in the terrain or the angle of the landslide mass.

Additionally, the interferometric images from GBInSAR are usually based on the relative changes in ground motion, while GNSS directly provides the absolute displacement values. This could also contribute to the difference in acceleration magnitudes between the two methods. GBInSAR is more dependent on displacement differences between adjacent time points, meaning its acceleration values may reflect relatively smaller variations, whereas GNSS can monitor absolute displacements in real time, especially during periods of significant displacement in the landslide area, when GNSS is often able to capture more noticeable acceleration changes.

Despite the differences in the specific magnitudes of acceleration, the acceleration curves from both methods show highly consistent trends during the accelerated deformation phase, indicating a synergistic effect between the two methods in landslide dynamic monitoring. This data comparison allows the conclusion that, although there are differences in the magnitude of acceleration between GNSS and GBInSAR, their acceleration change trends are consistent and mutually validating, providing reliable data support for precise landslide monitoring and early warning.

5.3 Limitations and challenges of multi-source ground monitoring methods

Although this study shows that integrating satellite-borne InSAR, GB-InSAR, GNSS, and rain-gauge observations can strengthen landslide monitoring and early warning, the main contribution of the framework is not simply generating more data, but improving reliability through cross-validation and enhancing process interpretability for precipitation-driven deformation. The co-registered time series presented in Figure 8 indicates that deformation accelerations and turning points are temporally aligned with concentrated rainfall episodes, while the agreement between GNSS and GB-InSAR during acceleration phases provides an independent constraint that helps avoid misinterpreting sensor-specific noise as true slope motion. This behavior aligns with the hydro-mechanical understanding described by Woods et al., 2021, in which rainfall infiltration modifies slope hydrological conditions and reduces material strength, thereby promoting deformation and instability. At the same time, the integrated observations help separate event-scale rapid responses from slower deformation under low rainfall, which has been discussed in the context of precipitation-controlled landslide evolution, including threshold and lag effects, by Schlögel et al. (2015), Yin et al. (2010), and Zhang et al. (2020).

Figure 8

Figure 8. Time series comparison of edge-radar displacement, GNSS displacement component dx, and rain-gauge precipitation.

Despite these advantages, several practical limitations remain. Satellite-borne InSAR is valuable for regional screening and hotspot identification, but its ability to capture short-lived accelerations is constrained by revisit interval, viewing geometry, and decorrelation in complex terrain. Consequently, it may not reflect precipitation-driven changes at the event scale in a timely manner, particularly for small or localized displacements. GB-InSAR provides dense, near-continuous measurements and is therefore well suited to capturing rainfall-triggered deformation dynamics, but it is a line-of-sight technique with limited areal coverage per deployment and can be affected by atmospheric phase delays and adverse weather during heavy rainfall. These effects may introduce short-term noise and require careful quality control and correction. GNSS offers high-precision three-dimensional deformation at discrete points and serves as a robust independent reference for validating radar-derived trends. However, its spatial representativeness is limited by sparse station distribution and potential signal obstruction in mountainous terrain, which can lead to data gaps or incomplete coverage of the moving body.

A further challenge lies in data fusion and interpretation. The four data streams differ in spatial support, temporal sampling, and reference frames, meaning that direct comparison requires strict time alignment, consistent referencing, and uncertainty-aware interpretation. Moreover, rain gauges provide point rainfall measurements that may not fully represent spatially heterogeneous precipitation over complex topography, which can complicate rainfall–deformation attribution at fine scales. Future work should therefore emphasize uncertainty propagation and reference unification across sensors, atmospheric and meteorological correction strategies for GB-InSAR during severe weather, optimized GNSS network placement to improve representativeness and robustness, and the use of spatial rainfall products where available to better characterize precipitation forcing. Overall, while each technique has inherent limitations, the proposed multi-source framework improves monitoring credibility by combining complementary strengths and providing cross-checked evidence for precipitation-driven landslide deformation.

6 Conclusion

This study demonstrates the significant correlation between precipitation events and landslide displacement, highlighting the direct impact of precipitation on landslide triggering. The results show that both short-duration intense rainfall and long-term accumulated precipitation are key factors in landslide occurrence, with accelerated displacement observed during high precipitation peaks. GB-InSAR and GNSS monitoring provided precise, high-frequency data that captured dynamic changes in the landslide area, revealing significant ground deformation. The fusion of these multi-source data methods validated the spatiotemporal relationship between precipitation and landslide displacement, offering reliable insights into the role of soil moisture and groundwater in landslide activity.

While multi-source data fusion proves valuable, limitations in spatial resolution, real-time capabilities, and monitoring coverage remain, particularly during the early stages of landslides. Future work should focus on optimizing data fusion techniques to enhance monitoring accuracy and mitigate the effects of meteorological and topographical influences on data quality.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Author contributions

DY: Writing – original draft, Methodology, Writing – review and editing, Formal Analysis. YD: Writing – review and editing, Project administration, Supervision, Writing – original draft. ZZ: Data curation, Writing – review and editing, Investigation, Writing – original draft.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This research was funded by Chengdu Municipal Science and Technology Program–Achievement Transformation Guidance PlanR&D and Application of an Integrated Space–Air–Ground Landslide Monitoring, Assessment, and Emergency Response SystemProject No.: 2023-CY02-00002-GX.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Abgrami, A., Zhang, W., Mao, H., and Wang, L. (2025). GIS-based comparative landslide susceptibility mapping for Kelardasht county with ANN, SVM and RF models. Civ. Geoengineering Lett. 2 (1), e100028. doi:10.22034/CGEL.2.1.e100028

CrossRef Full Text | Google Scholar

Acar, M., Ozludemir, M. T., Erol, S., Celik, R. N., and Ayan, T. (2008). Kinematic landslide monitoring with Kalman filtering. Nat. Hazards Earth System Sci. 8, 213–221. doi:10.5194/nhess-8-213-2008

CrossRef Full Text | Google Scholar

Ahadi, E., and Rosta, D. A. (2024). Landslide susceptibility analysis for Azershahr region using SVM and logistic regression methods. Civ. Geoengin. Lett. 1 (2), e100022.

Google Scholar

Birjandi, D., and Derakhshani, R. (2025). Landslide Hazard assessment induced by active Tectotics in the Golpayegan Region using AHP. Civ. Geoengin. Lett. 2 (2), e100043. doi:10.22034/CGEL.2.2.e100043

CrossRef Full Text | Google Scholar

Cai, J., Liu, G., Jia, H., Zhang, B., Wu, R., Fu, Y., et al. (2022). A new algorithm for landslide dynamic monitoring with high temporal resolution by Kalman filter integration of multiplatform time-series InSAR processing. Int. J. Appl. Earth Observation Geoinformation 110, 102812. doi:10.1016/j.jag.2022.102812

CrossRef Full Text | Google Scholar

Del Soldato, M., Confuorto, P., Bianchini, S., Sbarra, P., and Casagli, N. (2021). Review of works combining GNSS and InSAR in Europe. Remote Sens. 13, 1684. doi:10.3390/rs13091684

CrossRef Full Text | Google Scholar

Ebadati, M., Sun, F., Lee, Y., and Moghadam, M. J. (2025). Landslide susceptibility analysis using artifacial neuran netwonks for Chalus County, Iran. Civ. Geoengin. Lett. 2 (2), e100039. doi:10.22034/CGEL.2.2.e100039

CrossRef Full Text | Google Scholar

Ferrigno, F., Gigli, G., Fanti, R., Intrieri, E., and Casagli, N. (2017). GB-InSAR monitoring and observational method for landslide emergency management: the Montaguto earthflow (AV, Italy). Nat. Hazards Earth Syst. Sci. 17, 845–860. doi:10.5194/nhess-17-845-2017

CrossRef Full Text | Google Scholar

Gorsevski, P. V., and Jankowski, P. (2010). An optimized solution of multi-criteria evaluation analysis of landslide susceptibility using fuzzy sets and Kalman filter. Comput. Geosciences 36, 1005–1020. doi:10.1016/j.cageo.2010.03.001

CrossRef Full Text | Google Scholar

Hamedi, A., Nejhad, S. M., and Mofrad, H. H. (2025). AHP-based susceptibility analysis for Landslides in Fuman County, northwestern Iran. Civ. Geoengin. Lett. 2 (1), e100030. doi:10.22034/CGEL.2.1.e100030

CrossRef Full Text | Google Scholar

Karunathilake, A., Zou, L., Kikuta, K., Nishimoto, M., and Sato, M. (2019). Implementation and configuration of GB-SAR for landslide monitoring: case study in Minami-Aso, Kumamoto. Explor. Geophys. 50, 210–220. doi:10.1080/08123985.2019.1588069

CrossRef Full Text | Google Scholar

Liao, K., Zhang, W., Zhu, H., Zhang, Q., Shi, B., Wang, J., et al. (2022). Forecasting reservoir-induced landslide deformation using genetic algorithm enhanced multivariate Taylor series Kalman filter. Bull. Eng. Geol. Environ. 81, 104. doi:10.1007/s10064-022-02595-1

CrossRef Full Text | Google Scholar

Lingua, A., Piatti, D., and Rinaudo, F. (2008). Remote monitoring of a landslide using an integration of GB-INSAR and LIDAR techniques. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 37, 133–139.

Google Scholar

Liu, B., He, K., Han, M., Hu, X., Ma, G., and Wu, M. (2021). Application of UAV and GB-SAR in mechanism research and monitoring of Zhonghaicun landslide in Southwest China. Remote Sens. 13, 1653. doi:10.3390/rs13091653

CrossRef Full Text | Google Scholar

Lu, F., and Zeng, H. (2020). Application of Kalman filter model in the landslide deformation forecast. Sci. Rep. 10, 1028. doi:10.1038/s41598-020-57881-3

PubMed Abstract | CrossRef Full Text | Google Scholar

Moretto, S., Bozzano, F., and Mazzanti, P. (2021). The role of satellite InSAR for landslide forecasting: limitations and openings. Remote Sensing 13, 3735. doi:10.3390/rs13183735

CrossRef Full Text | Google Scholar

Mugnai, F., and Tarchi, D. (2022). Multiple-input multiple-output radar, ground-based MIMO SAR for ground deformation monitoring. Eur. J. Remote Sens. 55, 604–621. doi:10.1080/22797254.2022.2141660

CrossRef Full Text | Google Scholar

Qiu, Z., Jiao, M., Jiang, T., and Zhou, L. (2020). Dam structure deformation monitoring by GB-InSAR approach. Ieee Access 8, 123287–123296. doi:10.1109/access.2020.3005343

CrossRef Full Text | Google Scholar

Schlögel, R., Doubre, C., Malet, J.-P., and Masson, F. (2015). Landslide deformation monitoring with ALOS/PALSAR imagery: a D-InSAR geomorphological interpretation method. Geomorphology 231, 314–330. doi:10.1016/j.geomorph.2014.11.031

CrossRef Full Text | Google Scholar

Senogles, A., Olsen, M. J., and Leshchinsky, B. (2023). LADI: landslide displacement interpolation through a spatial-temporal Kalman filter. Comput. Geosciences 180, 105451. doi:10.1016/j.cageo.2023.105451

CrossRef Full Text | Google Scholar

Shi, M., Yang, H., Wang, B., Peng, J., Gao, Z., and Zhang, B. (2021). Improving boundary constraint of probability integral method in SBAS-InSAR for deformation monitoring in mining areas. Remote Sens. 13, 1497. doi:10.3390/rs13081497

CrossRef Full Text | Google Scholar

Su, Y., Yang, H., Peng, J., Liu, Y., Zhao, B., and Shi, M. (2022). A novel near-real-time GB-InSAR slope deformation monitoring method. Remote Sens. 14, 5585. doi:10.3390/rs14215585

CrossRef Full Text | Google Scholar

Tian, W., Zhao, Z., Hu, C., Wang, J., and Zeng, T. (2019). GB-InSAR-based DEM generation method and precision analysis. Remote Sens. 11, 997. doi:10.3390/rs11090997

CrossRef Full Text | Google Scholar

Wentao, Z., Wenjun, Z., Shihua, G., Yuntao, D., and Xianglong, L. (2021). Surface subsidence monitoring in mining area of SBAS-InSAR based on Sentinel-1A. J. Mines, Metals Fuels, 5–11. doi:10.18311/jmmf/2021/27316

CrossRef Full Text | Google Scholar

Woods, A., Macciotta, R., Hendry, M. T., Stewart, T., and Marsh, J. (2021). Updated understanding of the deformation characteristics of the checkerboard creek rock slope through GB-InSAR monitoring. Eng. Geol. 281, 105974. doi:10.1016/j.enggeo.2020.105974

CrossRef Full Text | Google Scholar

Yang, H., Liu, J., Peng, J., Wang, J., Zhao, B., and Zhang, B. (2020). A method for GB-InSAR temporal analysis considering the atmospheric correlation in time series. Nat. Hazards 104, 1465–1480. doi:10.1007/s11069-020-04228-w

CrossRef Full Text | Google Scholar

Yao, J., Yao, X., and Liu, X. (2022). Landslide detection and mapping based on SBAS-InSAR and PS-InSAR: a case study in Gongjue County, Tibet, China. Remote Sens. 14, 4728. doi:10.3390/rs14194728

CrossRef Full Text | Google Scholar

Yeganeh, A., Nazari, R., Zhu, D., Tagizadeh, P., and Chen, Z. (2025). Landslide inventory Databdse for Hazard Assessment in the Alborz Mountains, Iran. Civ. Geoengin. Letters 2 (1), e100035. doi:10.22034/CGEL.2.1.e100035

CrossRef Full Text | Google Scholar

Yin, Y., Zheng, W., Liu, Y., Zhang, J., and Li, X. (2010). Integration of GPS with InSAR to monitoring of the Jiaju landslide in Sichuan, China. Landslides 7, 359–365. doi:10.1007/s10346-010-0225-9

CrossRef Full Text | Google Scholar

Zhang, Y., Meng, X. M., Dijkstra, T. A., Jordan, C. J., Chen, G., Zeng, R. Q., et al. (2020). Forecasting the magnitude of potential landslides based on InSAR techniques. Remote Sens. Environ. 241, 111738. doi:10.1016/j.rse.2020.111738

CrossRef Full Text | Google Scholar