This study focuses on the ±800 kV Qishao Line transmission corridor, which traverses Badong County, located in the southwestern part of Hubei Province. The region is situated in the upper reaches of the Yangtze River, surrounded by the Xuefeng Mountains to the northwest and the Daba Mountains to the southeast. The study area spans diverse and complex terrain with an average elevation of about 800 meters. Badong’s geological conditions, including steep slopes and frequent geological hazards such as landslides and collapses, make it a critical area for infrastructure development, particularly for high-voltage transmission lines like the Qishao Line. This line plays a vital role in the energy transmission between Hubei Province and its neighboring regions, while also contributing to the region’s industrial development.

The terrain and geological features of the area, including sedimentary rocks and fault lines, require close monitoring to assess the stability of the transmission towers. Several sections of the Qishao Line, particularly in the mountainous regions, face challenges such as unstable slopes, rockfalls, and soil erosion, exacerbated by seasonal rainfall and human activities. The study of this transmission corridor highlights both the environmental challenges and the importance of infrastructure stability in this region in Figure 1. However, the complex geological environment and frequent meteorological disasters pose significant challenges to the stability of the transmission lines (Zhang et al., 2024).

Sentinel-1 is a radar satellite system under the European Union’s Copernicus Programme, operated by the European Space Agency (ESA). It is equipped with C-band synthetic aperture radar (SAR) capabilities, enabling high-resolution surface deformation monitoring under various weather conditions. Sentinel-1A was launched in 2014, and Sentinel-1B joined operations in 2016, together providing global coverage with a 6–12-day repeat cycle. The Sentinel-1 SLC (Single Look Complex) data used in this study were obtained from the ESA Open Data Hub (https://scihub.copernicus.eu/), covering the period from 29 July 2019, to 4 August 2020, with a total of 31 scenes. The satellite’s orbital coverage includes the UHV transmission corridor in the study area, providing high temporal and spatial resolution data support for InSAR deformation monitoring.

RADARSAT-2 is a commercial synthetic aperture radar satellite developed and operated by Canada’s MDA and the Canadian Space Agency. Since its launch in 2007, its multi-polarization and high-resolution imaging capabilities have been widely applied in geological disaster monitoring, environmental change tracking, and infrastructure risk assessment. The RADARSAT-2 data used in this study were obtained under license from MDA, covering the key transmission corridors in the research area. The data span from 14 August 2019, to 4 August 2020, with a total of 18 scenes. By combining RADARSAT-2 data with Sentinel-1 data, this study enhanced temporal monitoring density, improved identification accuracy, and addressed the limitations of using a single satellite, playing a crucial role in the overall analysis.

This study utilized Sentinel-1 C-band SAR data with VV polarization and both ascending and descending orbits, which offers wide coverage and a short revisit time (6 days), making it suitable for monitoring regional deformation trends. In contrast, RADARSAT-2 also operates in the C-band but typically uses HH polarization and a configurable beam mode (Fine Beam mode in this study), enabling higher spatial resolution and localized deformation detection.

Differences in polarization modes, imaging geometry, and orbital direction between Sentinel-1 and RADARSAT-2 may introduce systematic biases in areas with dense vegetation. Specifically, VV polarization tends to penetrate vertical structures more effectively, while HH polarization is more sensitive to surface roughness, potentially affecting coherence and phase response. To address these discrepancies, careful co-registration and normalization were performed during data preprocessing to reduce fusion-induced artifacts. In the result analysis, particular attention was paid to deformation anomalies in vegetated areas, which were evaluated in conjunction with field observations and auxiliary data to assess the potential bias introduced by sensor characteristics.

To evaluate the applicability and effectiveness of different SAR data in deformation monitoring, this study compares the InSAR monitoring results based on Sentinel-1 and RADARSAT-2 data, as shown in Figure 4. The figure presents the annual average deformation rate distribution along the ultra-high voltage transmission corridor, reflecting the overall surface deformation characteristics of the study area.

Figure 4a shows the deformation results based on Sentinel-1 data, which has a wide coverage and a dense distribution of deformation points. The annual average deformation rate in most areas of the study region falls between −10 mm/a and +10 mm/a, indicating that the transmission corridor is generally in a stable state. However, in localized areas, the deformation rate exceeds ±20 mm/a, suggesting the potential presence of geological hazard risks or ground subsidence issues. Due to the high temporal resolution and short revisit period of Sentinel-1 satellites, this dataset has advantages in extracting long-term, slow deformation trends.

Figure 4b presents the monitoring results based on RADARSAT-2 data. Although its coverage is slightly smaller than that of Sentinel-1, RADARSAT-2 data offers higher spatial resolution and demonstrates stronger detail recognition capability, especially in complex terrain areas such as slopes and valley intersections. As shown in the figure, the deformation patterns identified by RADARSAT-2 are generally consistent with those from Sentinel-1, further validating the reliability of the monitoring results. However, in certain localized areas with significant terrain undulations, RADARSAT-2 can more clearly identify small deformation regions, highlighting the advantage of its high-resolution C-band data in monitoring fine surface changes.To quantitatively evaluate the consistency between deformation results derived from Sentinel-1 and RADARSAT-2, we performed a pixel-wise comparison over their overlapping spatial coverage. The average annual deformation rates from both datasets were analyzed using several statistical metrics, including Pearson correlation coefficient (R), root mean square error (RMSE), and Spatial Coherence Index (SCI).

The analysis yielded a Pearson R value of 0.82, indicating strong positive correlation between the two datasets. The RMSE was calculated as 7.6 mm/yr, and the SCI reached 0.76, suggesting substantial spatial agreement. Despite inherent differences in sensor characteristics such as polarization, resolution, and orbit geometry, the results demonstrate significant consistency in the spatial pattern and deformation trends. Local discrepancies were primarily observed in vegetated or low-coherence zones, which aligns with the expected limitations of SAR-based measurements.

As shown in Figure 4, a notable discrepancy exists between the deformation rates derived from RADARSAT-2 (−95.6 mm/yr) and Sentinel-1 (−18.4 mm/yr) over the same region. This difference can be attributed to several key factors:

Firstly, the two datasets differ in polarization modes (HH vs. VV), spatial resolution, incidence angles, and orbit geometry, all of which can lead to systematic variations in coherence, phase stability, and displacement estimates. In particular, differences in polarization sensitivity under vegetated or rough surfaces may amplify displacement biases.Secondly, Sentinel-1 is optimized for wide-area monitoring and tends to yield more smoothed deformation patterns, while RADARSAT-2, operating in Fine Beam mode, offers higher spatial resolution and may better capture localized subsidence features.

Figure 5 shows the annual average surface deformation distribution for a typical slope area on the northern bank of the Yangtze River, based on Sentinel-1A (Figure 4a) and RADARSAT-2 (Figure 4b) data. The map background is a Google Earth optical image, providing a visual representation of the spatial distribution of typical subsidence areas. Both SAR datasets were processed using the SBAS-InSAR method, which optimized coherence through time-series interferometry and suppressed the impacts of spatial decorrelation and atmospheric errors. The resulting deformation data exhibit high consistency in their spatial and temporal distribution features.

In the Figure 5, areas of significant subsidence anomalies are shown in red, with subsidence rates generally exceeding −76 mm/yr, primarily concentrated in the northern bank region between 110°10′E and 110°11′E, extending in a strip-like pattern. This alignment is highly consistent with the slope direction of the bank and the flow direction of the river, reflecting a clear relationship between surface deformation and geomorphological units in the region.

Based on the area’s topography, geology, and human activity context, the subsidence deformation can be preliminarily attributed to a slip-subsidence composite mechanism dominating the slope deformation. The study area is a transitional zone of river valley terraces, characterized by steep slopes and surface cover primarily composed of Quaternary loose accumulation layers. The structure of the rock-soil body is loose, and its shear strength is relatively low, making it prone to shear softening and structural failure under river erosion. Furthermore, the deformation core area is near the construction zone of the riverbank and the road infrastructure along the river, suggesting that surface disturbances may have triggered shallow sliding or local instability processes.

The inversion results of Sentinel-1A and RADARSAT-2, which come from different orbital directions and polarization modes, are highly consistent in the subsidence anomaly areas, demonstrating the significant advantages of multi-source SAR in data complementarity and deformation recognition accuracy. This is especially true under conditions of high vegetation cover and complex terrain, where it effectively compensates for the limitations of single-satellite temporal and spatial resolution. In contrast, the southern bank region has gentler terrain, stable base lithology, and the monitoring results show relatively small overall deformation, with most annual average deformation values falling within ±5 mm/yr, indicating that the region remained relatively stable during the study period.

From a deformation mechanism perspective, the construction and operation of the transmission line corridor have had a significant impact on local slope stability. On one hand, to ensure the safety of the ultra-high voltage tower foundations, activities such as foundation reinforcement, pile foundation construction, and slope excavation are often required. These activities easily disturb shallow strata, damage the original stress field, and cause soil reorganization and a reduction in shear strength. On the other hand, long-term tower load accumulation and changes in surface drainage pathways may induce local stress concentration or water retention zones, thereby triggering secondary subsidence and creep deformation over time.

The deformation pattern along the northern bank of the Yangtze River shows characteristics of both progressive settlement and downslope sliding. Based on remote sensing results and field surveys, this study attributes the deformation to a “sliding–settlement compound mechanism.” Although no direct borehole or geotechnical test data were collected in this study, we reviewed publicly available geological reports, including the [YICHANG Geological Hazard Prevention Plan (2021–2025)]. These sources confirm the presence of soft to plastic silty clay layers (up to 20 m thick) in the area, which are prone to deformation under riverbank erosion and groundwater fluctuations.

In the localized high-subsidence zones (as shown in Figure 5), the deformation is preliminarily attributed to anthropogenic activities such as excavation and slope cutting. Although detailed construction records and timelines were not available in this study, visual interpretation of high-resolution remote sensing imagery and public planning documents indicate that large-scale site leveling and slope modification occurred in this area between 2019 and 2020. By overlaying spatial deformation rate maps with the temporal evolution of displacement, we observed that the onset of significant subsidence coincided with the estimated initiation of construction, followed by sustained deformation—suggesting a typical “disturbance-induced progressive settlement” pattern.

Therefore, in the absence of site-specific subsurface data, our interpretation is supported by deformation trends and regional geological information. We acknowledge this as a limitation and propose that future work should incorporate borehole drilling and stratigraphic profiles to validate the inferred compound mechanism.

Figure 6 presents the distribution of surface deformation features in a typical valley terrain area and their spatial relationship to the potential impacts on the ultra-high voltage (UHV) transmission line corridor. The deformation information is derived from InSAR inversion based on Sentinel-1A and RADARSAT-2 data (Figures 6a,b).

By comparing the InSAR deformation field with the geomorphological features, it is evident that two typical areas, outlined in white and blue, show significant differences in deformation magnitude and spatial distribution, reflecting the strong control of topographic factors on surface deformation evolution. The white-circled area, located at mid-to-low elevations, exhibits persistent subsidence characteristics, with the maximum annual average subsidence rate exceeding −20 mm/yr. The subsidence extent overlaps significantly with areas of human activity, indicating that the surface deformation in this region is largely controlled by engineering disturbances. The water accumulation effect at the foot of the slope and stress redistribution caused by construction loads may trigger shear deformation of shallow weak structures, leading to cumulative subsidence. This behavior shows an uneven deformation response driven by the coupling of engineering loading and geological background.

In contrast, the blue-circled area at higher elevations, situated in the ridge and steep slope transition zone, although having a greater slope and a certain degree of sliding potential under gravitational forces, shows relatively small deformation in the InSAR results, with annual average deformation values mostly below −5 mm/yr and a more dispersed spatial distribution. This phenomenon can be attributed to two factors: first, the surface in this area is predominantly original forest land, where the stratigraphic structure remains intact and the disturbance level is extremely low, so foundation settlement mechanisms have not been significantly triggered; secondly, SAR imagery in the high-slope areas exhibits strong geometric distortion effects, such as slope shielding and interference fringe breaks in shadowed areas. Overall, the valley topography directly affects the spatial distribution and deformation mechanisms of subsidence areas. Deformation accumulates in lower and middle elevations, while high and steep areas show either suppressed deformation or a stable state that has yet to significantly develop.

For transmission line corridors laid along valleys, subsidence in the low-lying sections may not only induce uneven tower foundation subsidence and foundation instability but may also be accompanied by risks of shallow shear sliding. Meanwhile, although the high-slope sections may remain stable in the short term, the potential risk of sudden landslides triggered by extreme weather events, such as heavy rain, should be considered.

Figure 6 illustrates the statistical distribution characteristics of surface deformation based on the InSAR inversion results from Sentinel-1A and RADARSAT-2 data. Figure 7a shows the deformation histogram from Sentinel-1A data, while Figure 7 shows the distribution of the RADARSAT-2 results. The red curve represents the normal distribution fitted to the standardized deformation rate, used to assess the central tendency and consistency of the deformation characteristics in both datasets.

The Sentinel-1A results show an approximately symmetric distribution, centered around zero but slightly skewed toward the negative, indicating minor localized subsidence in the study area. The deformation values primarily range from −30 mm/year to +20 mm/year, with most pixels having an annual average deformation rate within the −10 to +10 mm/year interval. This reflects a relatively stable overall deformation field in the region, with only a few areas exhibiting anomalous deformations.

In contrast, the RADARSAT-2-derived deformation values have a broader range, from about −50 mm/year to +100 mm/year, particularly showing a longer tail in the positive direction. This is likely due to its higher spatial resolution and greater sensitivity to engineering disturbances, especially in areas with significant topographical variations and intensive transmission line foundation construction, where larger uplift signals or error-amplified rapid deformations are more easily detected.Despite differences in observation resolution and imaging parameters between the two datasets, the overall deformation trends remain consistent, with both datasets showing an average deformation close to zero and no indication of large-scale systemic displacements. The normal fit curve shows that both datasets exhibit good concentration, further validating the comparability and integration value of multi-source InSAR data for regional surface deformation monitoring.

To assess whether the probability density functions (PDFs) of the Sentinel-1A and RADARSAT-2 datasets are consistent, we applied the Kolmogorov-Smirnov test and the Cramer-Von Mises test. Based on the results of these two tests, we confirm that the PDFs of the two datasets exhibit a high degree of consistency in deformation rate distribution characteristics. Specifically, the Kolmogorov-Smirnov test showed no significant statistical difference (p > 0.05), and the Cramer-Von Mises test also supported this conclusion, indicating that there is no significant difference in the distributions of the two datasets.

Before performing hotspot identification, we calculated the global Moran’s I index to assess the spatial autocorrelation of the deformation data. The result showed Moran’s I = 0.384 with p < 0.01, indicating significant positive spatial dependence and validating the suitability of applying local hotspot analysis using the Getis-Ord Gi* statistic.

To quantitatively assess the relationship between deformation hotspots and geological/engineering factors, we conducted chi-square tests and Spearman correlation analysis. Results indicated a significant association between hotspot areas and weak lithology (χ² = 15.72, p < 0.01), as well as high groundwater zones (ρ = 0.62, p < 0.01), confirming that the identified clusters are not only spatially significant but also geologically meaningful.

Figure 8 presents the surface deformation hotspot and coldspot spatial clustering analysis results based on InSAR technology. The Getis-Ord Gi* spatial statistical method was applied to grid the deformation values from the dual-source InSAR inversion, identifying areas of significant spatial clustering of surface subsidence (Hotspots) and relatively stable or upward-rebounding coldspot regions (Coldspots).

Hotspot areas are primarily concentrated along key engineering control sections of the northern bank of the Yangtze River, in low-lying or cut-slope excavation areas, exhibiting clear spatial continuity and clustering features. In contrast, coldspot regions are mostly located in high-elevation, undisturbed areas, such as forest-covered ridges or hilltops, where the deformation values are close to zero, and the spatial distribution is more dispersed.

In the hotspot areas, deformation values are generally greater than −15 mm/year and are distributed in a strip-like pattern along the engineering corridor. This indicates that these areas are not only controlled by natural terrain but are also strongly influenced by human engineering activities, including excavation, slope cutting, and tower foundation construction. Statistically, the z-score and p-value for these hotspots are significant (p < 0.05), confirming that they represent genuine deformation clustering rather than random distribution.

In contrast, the coldspot regions show stable deformation values with no significant clustering, likely corresponding to natural slope areas with stable geological conditions and no disturbance sources. The hotspot–coldspot spatial distribution pattern reveals the differentiated deformation evolution characteristics in the study area, reflecting the presence of potentially unstable sections and relatively safe sections within the engineering corridor. This has important implications for early warning systems and engineering management.

To further assess the consistency and complementarity of the deformation monitoring results from Sentinel-1A and RADARSAT-2, this study selected two sets of typical characteristic points and plotted their cumulative deformation time-series curves (Figures 7a,b).

Figure 9a shows the cumulative subsidence trends for characteristic points p1 and p2, derived from Sentinel-1A data. The results indicate that both points exhibit a clear downward trend. From July 2019 to September 2020, the cumulative subsidence for p1 was approximately −18.4 mm, while for p2, it was about −14.2 mm. The trends for both points are consistent, showing a stable subsidence evolution process with small fluctuations in the curve, resulting in an overall smooth dataset.

Figure 9b displays the corresponding time-series results from RADARSAT-2, showing larger subsidence magnitudes: p1 subsided approximately −95.6 mm, and p2 about −87.3 mm. The subsidence rates are higher than those obtained from Sentinel-1A, with more noticeable fluctuations in the curve. This is likely due to RADARSAT-2’s higher spatial resolution and enhanced response to local strong deformation regions under different imaging geometries. At the same time, it is important to note that the higher volatility might be influenced by factors such as sparse time-series data in small baselines and changes in coherence, suggesting that deformation extraction in high-dynamic areas requires careful management of coherence.

To further quantitatively assess the deformation consistency between the two data sources, Figure 9c presents a scatter density plot of deformation points across the entire region, with the x-axis representing the annual average deformation from RADARSAT-2, and the y-axis representing the corresponding values from Sentinel-1A. The color represents point density. The results show a high positive correlation between the two datasets, with the main distribution tightly clustered around the 45° reference line (y = x), indicating good consistency in the inversion results across most regions. The high-density core area in the density plot is primarily concentrated between −20 mm/year and +10 mm/year, reflecting that most surface deformations are in a state of mild subsidence or relative stability.

Overall, Sentinel-1A and RADARSAT-2 show strong consistency and complementarity in terms of deformation trends, spatial distribution, and statistical characteristics. Sentinel-1A is suitable for large-scale, long-term monitoring with stable inversion results, while RADARSAT-2 has an advantage in deformation sensitivity and detail performance in high-risk areas. The combined use of both datasets not only enhances the spatiotemporal coverage of deformation monitoring but also provides a data foundation for the engineering safety assessment and detailed management of high-risk areas.

Sentinel-1A data, with its high temporal resolution and stable revisit cycle, shows high stability and wide-area adaptability in capturing slow subsidence and long-term trends, enabling a comprehensive reflection of regional surface deformation field characteristics. The inversion results show that most of the transmission corridor area has a deformation rate within ±10 mm/a, with good data consistency, making it suitable for continuous monitoring and overall trend assessment. In contrast, RADARSAT-2, with its higher spatial resolution, demonstrates stronger local response capability in complex environments such as steep slopes, tower foundation construction areas, and valley terrains, providing more accurate identification of small subsidence and rapid deformation areas. The two datasets show good consistency in the time-series curves of characteristic points and deformation spatial patterns, with scatter density plot analysis showing a strong linear correlation between the two data sources, verifying the complementarity and fusion potential of multi-source InSAR data in regional infrastructure monitoring.

In the northern bank slope area of the Yangtze River channel, the monitoring results reveal a continuous subsidence zone, with local subsidence rates exceeding −76 mm/yr. The subsidence shows a strip-like concentration, closely related to the slope direction, geomorphological boundaries, and riverbank construction activities. Multi-source InSAR results consistently identify this as a typical subsidence hotspot, reflecting high instability risks due to the synergistic effects of riverbank erosion, infrastructure disturbance, and soft soil sliding, which pose a threat to the safety of nearby transmission tower foundations. Additionally, in the typical valley-crossing section, deformation hotspots are concentrated in the mid-to-low elevation zones, with significant subsidence at the slope foot, significantly influenced by water accumulation, construction loads, and weak structures. Although the high-slope sections have greater slope gradients, they show more stable deformation due to lower disturbance and good vegetation, appearing as coldspot regions. This spatial differentiation of deformation under the control of terrain height raises higher stability requirements for engineering in valley-crossing transmission line corridors.

Compared with previous studies that focused on urban or low-relief terrain, this research extends the application of multi-source InSAR to mountainous UHV transmission corridors. The integration of Sentinel-1A and RADARSAT-2, combined with spatial hotspot analysis, enables the identification of both regional deformation trends and localized high-risk areas. The observed differences between Sentinel-1 and RADARSAT-2 results can be largely attributed to their distinct sensor characteristics: Sentinel-1 operates in VV polarization with moderate spatial resolution and high revisit frequency, making it suitable for wide-area, long-term trend detection; while RADARSAT-2 employs HH polarization with higher spatial resolution and incidence angle variability, allowing it to capture localized, small-scale, and rapid deformations more sensitively—especially in rough terrain or construction-influenced zones. These differences underscore the value of combining sensors with complementary capabilities.

The main innovation of this study lies in the fusion-based interpretation of multi-source SAR time series in a geohazard-prone corridor, supported by deformation clustering and hotspot detection techniques. However, the absence of in situ validation data (e.g., GPS, leveling) and direct geological investigation (e.g., boreholes) limits the precision of causal inference. Future work should focus on integrating real-time ground-based monitoring data, high-resolution DEMs, and environmental parameters (e.g., rainfall, groundwater) to build dynamic deformation models and refine early warning systems. These improvements will facilitate intelligent and resilient management of critical transmission infrastructure in complex mountainous settings.