Time Series Analysis of Fucheng-1 Interferometric SAR for Potential Landslide Monitoring and Synergistic Evaluation with Sentinel-1 and ALOS-2

Highlights

What are the main findings?

• Time-series InSAR advantages: Fucheng-1 identified 13 small-scale potential landslides, whereas Sentinel-1 identified none, and achieved ~2.17× more identifications than ALOS-2. It also retrieved larger cumulative subsidence than Sentinel-1, driven by finer spatial sampling (higher spatial resolution) and a higher maximum detectable deformation gradient; this advantage becomes more evident as landslide size decreases.
• Interferometric performance and stability: Fucheng-1 provides 7–8× more high-coherence pixels than co-temporal Sentinel-1 and 1.1–1.4× more than ALOS-2, strengthening time-series inversion and enabling more spatially continuous deformation fields. Its orbital stability is comparable to Sentinel-1, and its maximum detectable deformation gradient in mountainous terrain is ~2× higher.

What are the implication of the main findings?

• Fucheng-1’s high deformation-identification sensitivity and high pixel density indicate its strong potential for landslide identification and hazard monitoring in complex terrain.
• Its stable interferometric quality and enhanced gradient detectability support its role as a valuable complement to existing SAR missions for regional and high-resolution deformation monitoring.

Abstract

Fucheng-1 is China’s first commercial synthetic aperture radar (SAR) satellite equipped with interferometric capabilities. Since its launch in 2023, it has demonstrated strong potential across a range of application domains. However, a comprehensive and systematic evaluation of its overall performance, including its time-series monitoring capability, is still lacking. This study applies the Small Baseline Subset (SBAS-InSAR) method to conduct the first systematic processing and evaluation of 22 Fucheng-1 images acquired between 2023 and 2024. A total of 45 potential landslides were identified and subsequently validated through field investigations and UAV-based LiDAR data. Comparative analysis with Sentinel-1 and ALOS-2 indicates that Fucheng-1 demonstrates superior performance in small-scale deformation identification, temporal-variation characterization, and maintaining a high density of coherent pixels. Specifically, in the time-series InSAR-based potential landslide identification, Fucheng-1 identified 13 small-scale potential landslides, whereas Sentinel-1 identified none; the number of identifications is approximately 2.17 times that of ALOS-2. For time-series subsidence monitoring, the deformation magnitudes retrieved from Fucheng-1 are generally larger than those from Sentinel-1, mainly attributable to finer spatial sampling enabled by its higher spatial resolution and a higher maximum detectable deformation gradient. Moreover, as landslide size decreases, the advantages of Fucheng-1 in deformation identification and subsidence estimation become increasingly evident. Interferometric results further show that the number of high-coherence pixels for Fucheng-1 is 7–8 times that of co-temporal Sentinel-1 and 1.1–1.4 times that of ALOS-2, providing more high-quality observations for time-series inversion and thereby supporting a more detailed and spatially continuous reconstruction of deformation fields. Meanwhile, the orbital stability of Fucheng-1 is comparable to that of Sentinel-1, and its maximum detectable deformation gradient in mountainous terrain reaches twice that of Sentinel-1. Overall, this study provides the first systematic validation of the time-series InSAR capability of Fucheng-1 under complex terrain conditions, offering essential support and a solid foundation for the operational deployment of InSAR technologies based on China’s domestic SAR satellite constellation.

1. Introduction

In recent years, the number of orbiting synthetic aperture radar (SAR) satellites has increased rapidly worldwide [1], providing robust hardware support for the study of geophysical processes [2,3,4] and the monitoring of geohazards [5,6,7,8]. Numerous commercial SAR systems have emerged globally, with companies such as Capella Space, Synspective (StriX), and Maxar launching high-resolution SAR missions that serve applications including urban deformation monitoring, resource assessment, and emergency response [9,10,11,12]. In contrast, China’s development of interferometric SAR (InSAR) missions began relatively late, with early research and operational applications relying heavily on foreign satellite data [13,14,15,16]. In recent years, however, the deployment of Chinese commercial SAR satellites has accelerated significantly, driven by organizations such as Xingluo, 4D TuXin, and China Siwei. These systems are now widely applied in geohazard monitoring, marine environmental surveillance, emergency response, urban safety, water resource management, and agricultural observation [17,18]. Nevertheless, many current commercial SAR satellites still lack stable interferometric capabilities, limiting their effectiveness in large-scale time series deformation analysis.

Fucheng-1 is China’s first commercial C-band synthetic aperture radar (SAR) satellite equipped with interferometric measurement capabilities. It was successfully launched by Spacety on 7 June 2023, marking the inaugural deployment of the Mianyang constellation. The satellite currently operates with an 11-day revisit cycle, which is expected to be further reduced in the future. Equipped with both left- and right-looking imaging modes, Fucheng-1 increases imaging frequency and provides diverse viewing geometries for observed targets. It supports high-resolution and wide-swath imaging, with scene widths ranging from 7 km to 170 km and spatial resolutions spanning from 20 m to as fine as 1 m × 0.5 m. Since its deployment, the satellite’s performance has garnered widespread attention within the remote sensing community. Building upon its initial success, Spacety subsequently launched Shenqi-01 and Shenqi-02 on 24 September 2024, and 17 May 2025, respectively, establishing a satellite constellation that significantly enhances both temporal resolution and overall system robustness.

InSAR and its time series extension (TS-InSAR) represent some of the most technically demanding yet operationally valuable applications of SAR technology, serving as key indicators of a satellite system’s capabilities. These methods are widely employed in ground deformation monitoring, geohazard early warning, and urban infrastructure stability assessment, and are widely accepted as essential for benchmarking the interferometric reliability of SAR platforms. Although Fucheng-1 has been utilized in preliminary studies such as mining subsidence [19,20] and landslide monitoring [21], existing research has primarily focused on single-pass interferometry (D-InSAR) and atmospheric correction techniques [22]. To date, no systematic assessment of Fucheng-1’s TS-InSAR capability has been conducted. Notably, Spacety is actively advancing its satellite constellation, highlighting InSAR as a strategic focus within its mission roadmap. Therefore, a comprehensive assessment of Fucheng-1’s time series interferometric capabilities is essential not only for supporting its operational deployment in large-scale geohazard monitoring, but also for promoting the independent advancement and engineering demonstration of China’s SAR systems in both InSAR and TS-InSAR domains.

This study provides the first comprehensive evaluation of Fucheng-1 for landslide monitoring, with a particular focus on time-series deformation and interferometric performance. The mountainous canyon region of Maoxian County, Sichuan Province, was selected as the study area. Using 22 scenes of Fucheng-1 C-band SAR imagery, we derived high-quality time series deformation results under complex terrain conditions. The accuracy and reliability of these deformation detections were validated through field investigations and UAV-based LiDAR data. In addition, through integration with Sentinel-1 and ALOS-2 datasets, we conducted a comparative evaluation of Fucheng-1’s performance across several key dimensions, including its ability to detect small-scale displacements, time series quality, interferometric stability, and maximum detectable deformation gradient (MDDG) in mountainous terrain.

2. Study Area and Datasets

2.1. Study Area

The study area is located in the northwestern part of Sichuan Province, southwestern China as shown in Figure 1. It lies within a subalpine monsoon climate zone, characterized by frequent torrential rainfall in spring and summer, and cold, dry conditions in winter. The regional topography generally slopes from northwest to southeast, with mountain peaks reaching elevations of approximately 4000 m and relative elevation differences ranging between 1500 and 2500 m. This dramatic terrain relief results in significant vertical and regional climatic differentiation, creating highly complex and variable local weather conditions [23,24,25]. Maoxian County, as one of the most geologically active regions in China is situated in the transitional zone between the Qinghai-Tibet Plateau and the Sichuan Basin. It is controlled by the intersection of multiple active fault zones and experiences frequent seismic activity. These tectonic conditions are recognized as a major triggering factor for the region’s high incidence of landslide hazards [14,26,27].

2.2. Datasets

This study utilized SAR datasets from three satellites: Fucheng-1, Sentinel-1 [28,29,30,31,32], and ALOS-2 [33,34,35], which include two C-band systems and one L-band system. The detailed parameters of these datasets are summarized in

Table 1. Summary of SAR Data Acquisition.

Parameter	Fucheng-1	Sentinel-1	ALOS-2
Spacing resolution (azimuth × range)	1.88 m × 1.25 m	13.95 m × 2.33 m	1.43 m × 2.12 m
Band	C	C	L
Wavelength(cm)	5.6	5.6	23.6
Orbit direction	Ascending	Ascending	Ascending
Acquisition period	Strip	TOPS	Strip
Acquisition period	9 November 2023–8 July 2024	16 August 2023–6 May 2024	26 November 2017–31 Mar 2019
Number of images	22	18	11

. Specifically, Fucheng-1 data span from 9 November 2023, to 8 July 2024, comprising 22 ascending-track scenes. Sentinel-1 data cover the period from 16 August 2023, to 6 May 2024, with a total of 18 scenes. The ALOS-2 dataset includes 11 scenes acquired between 26 November 2017, and 31 March 2019. These datasets were employed to systematically evaluate the time series monitoring performance of Fucheng-1 under complex mountainous terrain conditions. To correct for topographic phase components in the interferometric processing, a Shuttle Radar Topography Mission (SRTM) digital elevation model (DEM) with a spatial resolution of 30 m was used for topographic correction [36].

3. Methodology

This study evaluates the time series and interferometric performance of the Fucheng-1 satellite in steep mountainous regions, with a focus on its capability under complex terrain conditions. Comparative analyses were performed using Sentinel-1 and ALOS-2 datasets. As shown in Figure 2, the overall methodology is organized into three core modules:(I) Data processing, (II) Time Series for landslides monitoring, and (III) Evaluation of Fucheng-1 Characteristics.

Step I (Figure 2I) involves preprocessing of multi-source SAR data acquired from Fucheng-1, Sentinel-1, and ALOS-2. Supporting datasets include the Shuttle Radar Topography Mission Digital Elevation Model (SRTM-DEM) and precise orbit ephemerides. The GAMMA (v2022) software was used to complete a series of standard InSAR processing steps, including image co-registration, interferogram processing phase unwrapping, baseline refinement, and time series processing.

Step II (Figure 2II) focuses on the generation of high-quality time-series InSAR deformation results from Fucheng-1 in high-relief terrain. Based on these results, potential landslides were preliminarily identified. To verify the reliability of the identified deformation signals, a validation campaign was conducted using field investigations and UAV-based LiDAR data.

Step III (Figure 2III) presents a multi-dimensional assessment of Fucheng-1’s performance in mountainous environments, integrating results from all three SAR datasets. The evaluation includes four aspects: (a) advantage in identifying small-Scale deformations; (b) Time Series Deformation Analysis; (c) Interferometric Performance Evaluation; (d) Estimation of the maximum detectable deformation gradient (MDDG) under complex terrain conditions

3.1. SBAS-InSAR Method

Given that the objective of this study is to quantitatively evaluate the deformation monitoring capability of the Fucheng-1 satellite in time-series InSAR (TS-InSAR) applications, we employed the Small Baseline Subset (SBAS-InSAR) technique [37,38,39], which is theoretically mature and widely validated. This method was originally proposed by Berardino et al. in 2002 [40]. Compared with alternative InSAR approaches such as Distributed Scatterers (DS) [41], Persistent Scatterers (PS-InSAR) [42], and StaMPS [43], SBAS-InSAR was selected based on the following considerations:

(1) Standardized processing and greater comparability. SBAS-InSAR has a well-established and widely adopted processing workflow, whereas DS methods are highly dependent on algorithm design, software implementation, and operator experience. This study aims to assess the baseline performance of the Fucheng-1 satellite under its current imaging and data conditions. Therefore, more sophisticated processing methods that might artificially enhance the results were deliberately avoided to ensure that the evaluation reflects realistic performance under practical conditions.

(2) Improved adaptability to complex terrain and dense vegetation. In steep mountainous areas with substantial vegetation cover, temporal decorrelation is often pronounced. Building on the “small-baseline” concept, SBAS-InSAR [37,38,39] imposes strict constraints on both temporal and spatial baselines when constructing the interferogram network, prioritizing interferometric pairs with short temporal intervals and short spatial baselines. This strategy effectively enhances interferometric coherence, substantially reduces temporal decorrelation associated with increasing revisit intervals, and simultaneously mitigates geometric distortions and spatial decorrelation induces by variations in spatial baselines. Compared with PS-InSAR, which relies on a limited number of stable point targets, SBAS-InSAR better preserves the spatial continuity of deformation fields and is therefore more suitable for mountainous regions such as the study area.

To ensure parameter consistency and comparability across multi-source processing, all datasets were geometrically co-registered using a unified workflow, achieving a registration accuracy better than 1/8 of a pixel [44]. Multi-looking was applied with factors of 3 × 3 for Fucheng-1 and ALOS-2 and 4 × 1 for Sentinel-1. Phase filtering was performed with a uniform coherence threshold of 0.4 using a multi-scale, three-iteration window scheme, with window sizes decreasing from 128 × 128 to 64 × 64 and 32 × 32 pixels [45]. Phase unwrapping was conducted using the minimum cost flow (MCF) algorithm [46], with the same threshold of 0.4. To suppress atmospheric phase delays, an initial correction was applied via linear height fitting. Subsequently, recognizing that atmospheric delays exhibit high-pass behavior in time and low-pass behavior in space, and considering the spatiotemporal scales of the actual deformation field, temporal high-pass and spatial low-pass filtering were applied to the interferogram stack to further mitigate atmospheric artifacts. Finally, SBAS-InSAR time-series inversion was performed on the preprocessed datasets.

3.2. Index for Interferometric Performance Evaluation

The coherence of Interferometric Synthetic Aperture Radar (InSAR) is a fundamental parameter used to assess the quality of interferograms. It reflects the temporal stability of surface backscattering characteristics at the radar wavelength scale. According to [47], the total coherence can be decomposed into the product of several independent decorrelation components, expressed as:

$$\gamma ={\gamma }_{thermal}·{\gamma }_{baseline}·{\gamma }_{rotation}·{\gamma }_{volume}·{\gamma }_{temporal}·{\gamma }_{other}$$

(1)

These include thermal noise decorrelation, baseline decorrelation, spatial decorrelation, temporal decorrelation, and other system-induced decorrelation effects

Among them, thermal noise decorrelation (denoted as ${\gamma }_{thermal}$) arises from the inherent thermal noise associated with the signal-to-noise ratio (SNR) of the radar system [48]. The coherence due to thermal noise can be expressed as

$${\gamma }_{thermal}={\displaystyle \frac{1}{1+SN{R}^{-1}}}$$

(2)

The SNR itself is defined as:

$$SNR=10·\log \left({\displaystyle \frac{P_s}{P_n}}\right)$$

(3)

where $P_s$ is the average received signal power, $P_n$ is the thermal noise power of the receiving system, and $SNR$ is the signal-to-noise ratio of the radar system.

Baseline decorrelation is a form of geometric decorrelation that results from spatial differences between the master and slave SAR acquisitions. As the perpendicular baseline increases, geometric differences become more pronounced, reducing coherence. The baseline decorrelation can be approximated by:

$${\gamma }_{baseline}=\left\{\begin{array}{ll}1-{\displaystyle \frac{B_{\perp }}{B_c}},\hfill & B_{\perp }<B_c\hfill \\ 0,\hfill & B_{\perp }>B_c\hfill \end{array}\right.$$

(4)

The critical baseline $B_c$ which defines the threshold for coherence loss, is given by:

$$B_c={\displaystyle \frac{R{B}_w\tan ({\theta }_{inc}-\alpha )}{f}}$$

(5)

where $B_c$ is the critical baseline of the InSAR system, it depends on radar frequency ($f$), bandwidth ($B_w$), and topographic slope ($\alpha$).

The difference in Doppler centroid frequencies ($\Delta f_{DC}$) induced the geometric decorrelation in the azimuth direction, the coherence factor ${\gamma }_{DC}$ decreases linearly with increasing differences in the Doppler centroid frequencies, as follows:

$${\gamma }_{DC}=\left\{\begin{array}{ll}1-{\displaystyle \frac{\Delta f_{DC}}{B_A}},\hfill & |\Delta f_{DC}|\le B_A\hfill \\ 0,\hfill & |\Delta f_{DC}|>B_A\hfill \end{array}\right.$$

(6)

where $B_A$ is the bandwidth in the azimuth direction [44].

4. Results

4.1. Time Series Result of Fucheng-11 for Landslide Monitoring

Figure 3 presents the inverted annual-mean line-of-sight (LOS) velocity derived from Fucheng-1, together with the spatial distribution of the identified potential landslides. In total, 45 potential landslide sites were detected. As shown in the time series velocity result (Figure 3a), show no obvious elevation-correlated residuals or long-wavelength phase ramps, suggesting that major error components were effectively mitigated and that the inversion is reliable. Benefiting from the high spatial resolution of Fucheng-1, the results contain a high density of valid pixels, substantially enhancing the spatial granularity of deformation characterization. Deformation features are clearly delineated with well-defined boundaries.

The spatial distribution of the detected potential landslide indicates that most are located along national highways, in close proximity to several residential settlements. Should slope failures occur, they may pose serious threats to local populations and infrastructure. Figure 3d illustrates a landslide cluster with typical deformation patterns, where multiple sliding units exhibit distinct InSAR signals. Corresponding UAV-based LiDAR data (Figure 3e) reveal prominent rear scarp cracks, which align closely with the InSAR observations, further validating the accuracy of the deformation detection.

4.2. Field Verification

To further validate the potential landslides identified by Fucheng-1, we conducted field verification at representative sites marked in Figure 3(b1,b2,c1,c2,d1), with the corresponding results presented in Figure 4. At several key locations, UAV-borne LiDAR data were additionally incorporated (see Figure 4j,k) to improve the accuracy assessment and to support a detailed interpretation of surface-deformation features.

At the b1 landslide site (Figure 3(b1)) in Huilong Township, the Fucheng-1 time-series InSAR results indicate a deformation rate of −122 mm/yr (Figure 3b), suggesting active and relatively rapid downslope movement. Field investigations revealed pronounced surface cracks along the road at the slope toe (Figure 4c,d), consistent with the InSAR-derived deformation signal. Similarly, the b2 landslide shows sustained deformation in the InSAR observations, and on-site inspection documented clear evidence of intense sliding activity, including a visibly loosened slope surface and a well-defined sliding zone (Figure 4a). For sites c1 and c2, field verification indicates that c1 is characterized by extensively fragmented slope materials and obvious signs of recent landslide activity (Figure 4g), whereas c2 exhibits multiple surface cracks along the roadside (Figure 4i), implying ongoing active deformation. Site d1 was assessed by integrating UAV-LiDAR observations (Figure 4j,k) with field evidence (Figure 4b,e): rear-scarp cracks are clearly identifiable in the UAV imagery, and field observations further confirm persistent movement in the upper part of the slope with a well-defined landslide boundary. Deformation-related road damage is also documented in Figure 4f,i, demonstrating the impact of slope instability on local infrastructure.

Overall, the field surveys and UAV-LiDAR analyses provide strong, independent evidence supporting the reliability and accuracy of the Fucheng-1 time-series InSAR results. In total, five sites were validated through on-site inspections, and an additional seven sites were confirmed using UAV-based LiDAR data (Figure 4j,k). All validated locations are consistent with the identified potential landslide hazards, providing independent support for the reliability of our identification results. These findings demonstrate the strong capability and application potential of Fucheng-1 for potential landslide identification and operational geohazard monitoring.

5. Discussion

5.1. Small-Scale Deformation Detection Capability

To evaluate the deformation detection performance of Fucheng-1, a systematic assessment was conducted based on both original interferometric and time series results. Figure 5 presents a comparative analysis of interferograms from Fucheng-1, Sentinel-1, and ALOS-2 over a small-scale landslide-prone area. The specific interferometric pair information for each satellite is summarized in

Table 2. Acquisition and Baseline Parameters of Three Representative Interferometric Pairs.

Parameter	Fucheng-1	Sentinel-1	ALOS-2
Date of acquisition	25 January 2024–9 March 2024	31 January 2024–19 March 2024	26 November 2017–24 December 2017
Perpendicular baseline(m)	−37.94 m	−31.45 m	−212.99 m
Time baseline(days)	44	48	42
Spacing resolution (azimuth × range)	1.88 m × 1.25 m	13.95 m × 2.33 m	1.43 m × 2.12 m

.

The potential landslide deformation zone covers approximately 95 m × 210 m, representing a typical small-scale, localized deformation unit. High-resolution UAV-LiDAR imagery clearly reveals surface cracks and other deformation-related morphological features at the corresponding location (Figure 5j), providing independent external evidence to support the InSAR-derived identification.

Among the three datasets, Fucheng-1 exhibits the clearest and most spatially continuous interferometric fringes over this localized deformation area (Figure 5a). Under a consistent processing workflow and parameter configuration applied to all datasets (identical coherence thresholds and filtering strategy), the Fucheng-1 interferogram retains high fringe clarity and structural integrity after moderate filtering (coherence threshold of 0.4; 64 × 64 window) (Figure 5d), indicating stable spatial-structural characteristics of the signal. Furthermore, the line-of-sight (LOS) deformation field obtained via minimum-cost flow (MCF) phase unwrapping under a coherence constraint of 0.4 (Figure 5g) shows a continuous and concentrated localized deformation pattern and resolves multiple relatively independent sliding sub-units, whose spatial distribution is consistent with the deformation extent indicated by the LiDAR observations.

By contrast, Sentinel-1 shows no stably identifiable localized deformation response in either the interferograms or the LOS results for this area (Figure 5b,e,h). ALOS-2 exhibits only weak deformation signatures in the raw interferogram (Figure 5c), which are further smoothed and attenuated after filtering (Figure 5f); this behavior is jointly related to its L-band characteristics and the geometric configuration of the interferometric pair used in this analysis.

Quantitative profile comparisons further support the above assessment. Within the 95–210 m interval, the Fucheng-1 profile displays a pronounced deformation trough with a peak displacement of approximately −25 mm (Figure 5k), enabling clear characterization of the local deformation gradient. In contrast, the Sentinel-1 and ALOS-2 profiles show limited overall variation and relatively flat curves, indicating insufficient sensitivity to this small-scale localized deformation. Considering the interpretability of the interferometric fringes, the continuity of the unwrapped deformation field under a uniform threshold, the quantitative profile differences, and the independent LiDAR evidence, we conclude that Fucheng-1 performs better in detecting small-scale surface displacements and characterizing localized deformation patterns.

To comprehensively evaluate the potential of Fucheng-1 for deformation monitoring, this study conducted a comparative analysis under a unified InSAR processing framework. Time series deformation results were generated using 18 scenes of Sentinel-1 data (16 August 2023–6 May 2024) and 11 scenes of ALOS-2 data (26 November 2017–31 March 2019) over the same study area. Based on this analysis, the time series monitoring capability of Fucheng-1 was systematically compared with that of the other sensors. As shown in Figure 6, Fucheng-1 identified a total of 45 potential landslides, outperforming Sentinel-1 (20 detections) and ALOS-2 (32 detections), and demonstrating the highest detection capability among the three.

To further investigate the sensitivity of Fucheng-1 across different potential landslide scales, a three-level classification system was established based on potential landslide area: Small areas (<0.1 km²), Medium areas (0.1–0.2 km²), and Large areas (>0.2 km²). Statistical results for each class are illustrated in Figure 6d.

In the case of small-area landslides (<0.1 km²), Fucheng-1 successfully detected 13 potential events, markedly outperforming ALOS-2 (6 cases) and Sentinel-1 (0 cases). This superior performance is attributed to two primary factors. First, Fucheng-1 operates in the C-band, which has a shorter wavelength than the L-band of ALOS-2, thereby providing greater phase sensitivity to small displacements under equivalent deformation conditions. Second, the higher spatial resolution of Fucheng-1 enables a denser distribution of coherent pixels and finer spatial sampling, which facilitates the detection of localized deformation patterns in small-area landslides.

For medium-area landslides (0.1–0.2 km²), Fucheng-1 identified 9 cases, compared to 5 cases detected by both ALOS-2 and Sentinel-1. This further demonstrates Fucheng-1’s advantage in spatial resolution and coverage, particularly for moderate terrain units.

In large-area landslides (>0.2 km²), the number of detected events was relatively comparable across sensors: 23 for Fucheng-1, 21 for ALOS-2, and 15 for Sentinel-1. The variation in detection capability at this scale is primarily influenced by sensor wavelength characteristics—whereas the L-band offers deeper penetration and performs well in densely vegetated areas, the C-band is more sensitive to near-surface deformations.

It is worth noting that, judging from the visual appearance of the temporal maps, Sentinel-1 may exhibit fewer missing values and a more visually continuous pattern. This is mainly attributable to the strong filtering framework adopted in this study (a coherence threshold of 0.4, together with a three-stage iterative window filtering scheme using 128 × 128, 64 × 64, and 32 × 32 windows). Under this framework, low-coherence areas in lower-spatial-resolution data are more readily smoothed by filtering, thereby appearing more continuous in the maps. By contrast, higher-resolution data provide stronger capability for detail delineation and boundary preservation, and thus tend to retain true spatial discontinuities and their edges, potentially resulting in sharper missing-value boundaries in the maps. This discrepancy is a reasonable outcome arising from differences in spatial resolution and filtering response, and it also indirectly highlights the advantage of Fucheng-1 in pixel-level characterization and information preservation.

In summary, Fucheng-1 exhibits heightened sensitivity and precision in detecting small-area landslides and capturing subtle surface displacements, owing to its higher spatial resolution and the intrinsic phase sensitivity of the C-band. It detected more than twice the number of small landslides compared to ALOS-2 (13 vs. 6), and significantly outperformed the lower-resolution C-band Sentinel-1, which detected none. The increased density of coherent scatterers and finer spatial granularity allow Fucheng-1 to clearly delineate subtle deformation gradients and multi-tiered sliding zones. These capabilities are critical for the early identification and fine-scale delineation of geohazard-prone areas, underscoring the satellite’s strong potential for operational landslide monitoring.

5.2. Time Series Deformation Analysis

Figure 7 presents a comparison of time-series InSAR deformation results derived from Fucheng-1, co-temporal Sentinel-1, and ALOS-2 for four representative potential landslides. Overall, Fucheng-1, benefiting from its high spatial resolution, demonstrated superior performance across three key dimensions: pixel density, deformation rate sensitivity, and internal structural detail resolution.

First, the number of valid pixels extracted by Fucheng-1 in potential landslides were 195,657, 82,935, 36,603, and 16,391, respectively approximately 5 to 7 times higher than those of Sentinel-1 and 1.1 to 1.4 times greater than those of ALOS-2. This substantial increase in pixel density improved the completeness and clarity of the deformation fields, enabling the detection of small-scale gradient structures such as multi-tiered sliding zones. Such detail is critical for the accurate delineation of landslide boundaries and for interpreting the internal structure of slope failures.

Second, a comparison of co-temporal C-band datasets revealed that the median annual subsidence rates extracted by Fucheng-1 were −100, −95, −100, and −80 mm/year for the four sites, significantly exceeding those observed by Sentinel-1 (−50, −45, −20, and −20 mm/year). These represent improvements of approximately 2 to 5 times. Furthermore, Fucheng-1 captured more pronounced negative tail values, 50 to 100 mm/year deeper, demonstrating greater sensitivity to deformation. Although the ALOS-2 dataset was not temporally aligned with the C-band acquisitions, Fucheng-1 still reconstructed deformation fields with more complete patterns and sharper boundaries due to its finer spatial sampling.

From a spatial-scale perspective, Figure 7(a1–d1) shows that the areas of the four representative deformation zones decrease progressively. Fucheng-1 consistently retrieves continuous deformation signals with well-defined boundaries for both large-area deformation (a1,b1) and small-area deformation (c1,d1). In contrast, as the deformation area decreases, the deformation signature in Sentinel-1 becomes increasingly weak, and almost no effective signal is discernible for the fourth small-scale deformation zone (d2). This discrepancy is mainly attributed to the higher spatial resolution of Fucheng-1 (1.88 m × 1.25 m). Its finer pixel sampling effectively mitigates pixel-mixing and spatial-averaging effects, thereby more clearly resolving localized deformation that tends to be diluted or even obscured in the coarser-resolution Sentinel-1 data (13.95 m × 2.33 m). Importantly, this advantage in detecting small-scale deformation is not limited to single interferometric pairs (e.g., Figure 5) but becomes more pronounced after time-series InSAR inversion. The denser set of valid pixels provided by Fucheng-1 supplies more high-quality observations for spatiotemporal filtering and least-squares inversion, enabling noise suppression while preserving genuine localized deformation signals. By comparison, in small-scale deformation zones, Sentinel-1 often suffers from insufficient high-coherence observations and a lower signal-to-noise ratio, causing spatiotemporal filtering and inversion to further smooth or suppress subtle deformation, which ultimately hinders stable detection.

The displacement time series (Figure 7(a5–d5)) further corroborate the aforementioned differences in deformation rates and help explain their origin from a temporal-evolution perspective. Over the full time span, Fucheng-1 accumulates larger subsidence displacements within a shorter observation window; consequently, the inferred deformation rates from Fucheng-1 are overall markedly higher than those from Sentinel-1, consistent with the rate-based statistical comparisons. Focusing specifically on the temporally overlapping interval, the two pronounced-deformation sites (Figure 7(a5,b5)) exhibit highly consistent displacement trends and phase evolution between Fucheng-1 and Sentinel-1, indicating that Fucheng-1 provides time-series detection performance comparable to, or even better than, Sentinel-1 under clear deformation signals (with similar trends but larger cumulative subsidence magnitudes from Fucheng-1). In contrast, for the small-deformation sites (Figure 7(c5,d5)), Fucheng-1 still robustly resolves progressive subsidence and subtle temporal fluctuations, whereas the Sentinel-1 time series shows only limited variability, suggesting substantially lower sensitivity to small-magnitude deformation than Fucheng-1.

It is noteworthy that the Fucheng-1 and ALOS-2 acquisitions are not temporally coincident; therefore, we do not perform a direct comparison of deformation-rate magnitudes or time-series evolution between the two datasets, so as to avoid potential misinterpretation arising from temporal inconsistency. Instead, our comparative analysis focuses on metrics that are less sensitive to acquisition-time mismatch and are more indicative of landslide-monitoring capability, including landslide boundary delineation (Figure 7), with Figure 8 further emphasizing the co-temporal comparison between Fucheng-1 and Sentinel-1; as well as orbit-control performance (Figure 9), and the maximum detectable deformation gradient (Figure 10 and Figure 11). In terms of boundary mapping, both Fucheng-1 and ALOS-2 clearly delineate landslide margins and overall outperform Sentinel-1, which we attribute in part to their finer spatial resolutions (Fucheng-1: 1.88 m × 1.25 m; ALOS-2: 1.43 m × 2.12 m) relative to Sentinel-1 (13.95 m × 2.33 m), enabling sharper and more reliable boundary identification.

Although Fucheng-1 performs well in detecting small-scale deformation and resolving fine structural details, it still has non-negligible limitations for engineering-oriented and large-area operational applications. First, in terms of spatial coverage, the swath width of a single Fucheng-1 scene is approximately 7–170 km. Compared with the typical wide-area coverage of Sentinel-1 (250 km × 250 km), Fucheng-1 generally requires more tracks and/or more scenes to mosaic an equivalent area for rapid large-scale mapping, which increases mosaicking-consistency requirements and processing burden, and may reduce the timeliness and cost-effectiveness of regional monitoring. Second, with respect to data availability, Fucheng-1 was launched in 2023, and the historical archive currently remains limited for long-term analyses. In contrast, Sentinel-1 has a longer in-orbit record and a much larger data archive, providing stronger support for multi-year time-series InSAR analyses. From an operational perspective, Fucheng-1 is better suited to applications such as detailed monitoring of prioritized hazards and capturing rapid deformation over small areas, whereas Sentinel-1 is more appropriate for wide-area continuous coverage, regional inventory mapping, and long-term monitoring.

5.3. Interferometric Analysis over Steep Mountainous Regions

To comprehensively assess the interferometric performance of Fucheng-1 under complex terrain conditions, Sentinel-1 and ALOS-2 datasets covering the same area were selected as comparative benchmarks (see

Table 3. The Coherence Parameter Between Three Datasets.

Parameter	Fucheng-1	Sentinel-1	ALOS-2
Date of acquisition	20231120–20231201/ 20231201–20231223/ 20240114–20240216	20231120–20231202/ 20231202–20231226/ 20240107–20240212	20171126–20171224/ 20171224–20180204/ 20171126–20180204
NESZ(SNR−1,db)	<−22	−21.3	<−29
Coherence	0.3/0.29/0.3	0.4/0.39/0.41	0.43/0.39/0.4
Pixels with coherence > 0.5	29.2 million/ 27.9 million/ 30.6 million	3.7 million/ 3.6 million/ 3.8 million	21.9 million/ 18.1 million/ 29.2 million
Perpendicular baseline(m)	81/−77/−31	47/199/−37	0.54/−212/−212
Time baseline(days)	11/22/32	12/24/36	28/42/70
Wavelength(cm)	5.56	5.55	12.6
Spacing resolution (azimuth × range)	1.88 m × 1.25 m	13.95 m × 2.33 m	1.43 m × 2.12 m

). To preserve the native imaging characteristics of each dataset as much as possible, a uniform 1:1 multilooking strategy was employed during processing, and coherence was estimated using a 5 × 5 pixel window. It is important to note that due to differences in imaging modes and resolutions across sensors, their point spread functions (PSFs) and ground pixel representations differ. Therefore, the analysis in this section emphasizes practical applicability for engineering purposes rather than aiming for “absolute comparability” among imaging modes.

From the perspective of interferogram patterns (Figure 8(a2–j2)), Fucheng-1 and Sentinel-1 exhibit high consistency, indicating comparable interferometric capabilities. In contrast, ALOS-2, leveraging its L-band wavelength, offers inherent advantages in mitigating volumetric scattering and temporal decorrelation. Even with a temporal baseline of 70 days and a perpendicular baseline of $B_{\perp }$ ≈ −212 m, it maintains an average coherence of approximately 0.4 (see Figure 8 and Table 3), demonstrating superior interferometric stability in mountainous areas.

Under comparable temporal and perpendicular baselines, Fucheng-1 exhibits slightly lower average coherence than Sentinel-1. To further investigate the underlying causes, various sources of decorrelation were analyzed based on the total coherence model (Equation (1)). The model indicates that coherence is jointly influenced by thermal noise, baseline decorrelation, spatial decorrelation, temporal decorrelation, and system-related errors. According to the NESZ (noise equivalent sigma zero) values of the three satellites, all are below −22 dB, suggesting similar thermal noise levels with negligible impact on overall coherence, although ALOS-2 shows a slight advantage.

Given that the temporal and perpendicular baselines of the Fucheng-1 and Sentinel-1 interferometric pairs are comparable, their temporal and baseline decorrelation effects should, in principle, be essentially equivalent. According to the coherence model, the influence of thermal noise, Doppler centroid offsets, and orbit errors on the overall coherence can be considered negligible. The primary factor responsible for the observed difference in mean coherence is more likely attributable to disparities in imaging resolution. Specifically, Fucheng-1 operates in a high-resolution stripmap mode, where the smaller ground sampling cell and more discrete distribution of scatterers make spatial decorrelation more pronounced. In contrast, Sentinel-1 employs a lower-resolution TOPSAR mode with a larger equivalent pixel footprint, which enhances the statistical averaging effect and consequently yields higher coherence. Secondary contributions to coherence differences may also arise from variations in imaging mode and system implementation, including radar architecture, antenna pointing stability, and attitude control performance.

As shown in Table 3, the number of pixels with coherence greater than 0.5 in Fucheng-1 interferograms is 29.2 million, 27.9 million, and 30.6 million, respectively, while for Sentinel-1, the corresponding numbers are only 3.7 million, 3.6 million, and 3.8 million. The number of high-coherence pixels in Fucheng-1 is approximately 7 to 8 times that of Sentinel-1 and also 1 to 1.5 times higher than ALOS-2. This advantage provides a richer and higher-quality set of persistent scatterers for subsequent InSAR time-series analysis, significantly enhancing the accuracy and spatial continuity of ground deformation retrieval. As such, Fucheng-1 demonstrates strong potential and promising prospects for high-resolution, fine-scale ground deformation monitoring in practical engineering applications.

The capability of orbit control directly affects the interferometric performance of SAR imagery. In this study, the long-term orbital state vectors of 22 scenes acquired by Fucheng-1 were comparatively analyzed alongside those of Sentinel-1 and ALOS-2 to assess their respective orbital stability. The detailed results are presented in Figure 9. The findings indicate that the orbital trajectories of Fucheng-1 and Sentinel-1 exhibit a high degree of consistency over time, demonstrating excellent orbital overlapping and thus reflecting superior orbital stability. In contrast, ALOS-2 shows a relatively lower degree of orbital consistency, characterized by noticeable trajectory deviations. This discrepancy is largely attributed to ALOS-2’s design philosophy, which incorporates a longer temporal baseline and thereby allows greater orbital tolerance. While such a design enhances sensitivity to topographic variations, it also introduces increased orbital variability. In terms of orbital precision control, however, Fucheng-1 exhibits more robust performance, which contributes significantly to its outstanding interferometric capability.

5.4. Maximum Detectable Deformation Gradient in Mountainous Areas

In practical applications, when the ground deformation gradient exceeds the maximum detectable deformation gradient (MDDG) permitted by InSAR technology, the resulting interferogram fails to accurately reflect actual surface deformation patterns [49]. This issue is particularly prominent in mountainous regions, where complex terrain significantly influences the MDDG. Variations in slope and aspect can lead to notable changes in ground resolution, thereby affecting the satellite’s capability to detect surface deformation.

To systematically evaluate the applicability of the Fucheng-1 satellite for landslide monitoring in mountainous areas, three representative slopes (labeled 1, 2, and 3) were selected, each characterized by distinct combinations of slope steepness and aspect. A comparative analysis of the MDDG values was conducted using SAR datasets from Fucheng-1, Sentinel-1, and ALOS-2 (as illustrated in Figure 10). The results reveal that ALOS-2 demonstrates the highest MDDG, with values of 34.26 mm/m, 61.41 mm/m, and 33.80 mm/m across the three slopes, respectively, underscoring its pronounced advantage in monitoring landslides involving large deformation gradients. Fucheng-1 ranks second, with MDDG values of 6.35 mm/m, 20.32 mm/m, and 5.56 mm/m, reflecting its solid capability for detecting moderate deformation gradients. Sentinel-1 exhibits the lowest MDDG values, measured at 3.09 mm/m, 11.10 mm/m, and 2.97 mm/m, respectively, indicating its limited adaptability in regions with intense deformation. These findings highlight Fucheng-1’s favourable performance in mountainous areas subject to moderate deformation gradients and reaffirm ALOS-2’s superior suitability for monitoring high-gradient deformation scenarios. The results provide a reliable reference for selecting appropriate satellite data in engineering applications.

Figure 11 presents the maximum detectable deformation gradients (MDDG, in mm/m) for the three SAR satellites. ALOS-2 demonstrates the highest MDDG value, reaching 82 mm/m, which highlights its strong capability in monitoring large-magnitude landslide deformation. Fucheng-1 achieves an MDDG of 23 mm/m, indicating a robust performance in detecting moderate to significant deformation signals. In contrast, Sentinel-1 exhibits a comparatively lower MDDG of 12 mm/m, suggesting limited sensitivity to intense surface displacement.

Although both Fucheng-1 and Sentinel-1 operate in the C-band, Fucheng-1’s higher spatial resolution significantly enhances its sensitivity to deformation, underscoring the critical influence of both radar wavelength and spatial resolution in determining MDDG performance. Among the three datasets, ALOS-2 is best suited for detecting large-scale, fast-moving landslides due to its high MDDG. Notably, Fucheng-1 outperforms Sentinel-1 in large deformation detection despite sharing the same radar band, achieving an MDDG value nearly twice as high. This advantage stems from Fucheng-1’s finer spatial resolution, which not only improves its capacity to capture pronounced deformation but also enables the identification of smaller, potentially hazardous slope movements. Its superior balance between spatial resolution and deformation gradient sensitivity makes Fucheng-1 a more adaptable and effective option for InSAR-based geohazard monitoring across a wide range of terrain and deformation scenarios.

6. Conclusions

As China’s first high-resolution commercial SAR satellite with interferometric capabilities, Fucheng-1 represents a milestone in the development of China’s SAR technology. While Fucheng-1 has achieved success in various applications, TS-InSAR applications remain relatively underexplored. In this study, high-quality time-series InSAR results were obtained from Fucheng-1 over a mountainous canyon region. A total of 45 potential landslides were identified, and representative sites were validated through field investigations and UAV-based LiDAR data. Further comparative analyses with Sentinel-1 and ALOS-2 enabled a comprehensive evaluation of the time-series monitoring and interferometric capabilities of Fucheng-1. The main conclusions are as follows:

(1) Time-Series Quality and Validation: Fucheng-1 provided reliable, high-quality time-series InSAR results in steep mountainous terrain. A total of 45 potential landslides were identified, and representative sites were validated using field surveys and UAV-based LiDAR data, supporting the reliability of the monitoring results.

(2) Identification of Small-Scale Potential Landslides: Fucheng-1 demonstrated excellent capability in identifying localized deformation in both single interferograms and time-series results. It successfully characterized a small-scale deformation feature of ~95 m × 210 m with an LOS displacement of approximately −25 mm, reflecting high spatial resolution and strong interferometric performance. For small-area potential landslides, Fucheng-1 identified 13 cases, whereas Sentinel-1 identified none, and it achieved ~2.17× more identifications than ALOS-2, underscoring its sensitivity to subtle deformation in complex topography.

(3) Time-Series Monitoring Characteristics: The number of valid observation pixels extracted by Fucheng-1 in landslide-prone areas was ~5–8× that of Sentinel-1 and ~1.1–1.4× that of ALOS-2. Under the same processing conditions, the absolute magnitudes of the annual mean deformation rates retrieved from Fucheng-1 are approximately 2–5× larger than those from Sentinel-1. This advantage is mainly attributable to the high spatial resolution of Fucheng-1, which enables denser spatial sampling, more accurate boundary delineation, and improved identification of subtle internal slope deformation.

(4) Interferometric Performance: The overall interferometric performance of Fucheng-1 is comparable to that of Sentinel-1, whereas ALOS-2 maintains higher coherence in mountainous regions due to its longer L-band wavelength. Analysis based on the total-coherence model suggests that coherence differences between Fucheng-1 and Sentinel-1 may be related to factors such as imaging resolution and acquisition mode. Nevertheless, Fucheng-1 yields substantially denser high-coherence observations (e.g., coherence > 0.5) than Sentinel-1 (and is comparable to or higher than ALOS-2), which strengthens time-series inversion and improves the spatial continuity and accuracy of deformation retrieval. Orbit-stability assessments further indicate that the positioning accuracy of Fucheng-1 is comparable to Sentinel-1 and higher than ALOS-2, meeting reliability requirements for long-term interferometric monitoring.

(5) MDDG Capability in Mountainous Regions: Under complex mountainous terrain conditions, Fucheng-1 achieved an MDDG of approximately 23 mm/m, nearly twice that of Sentinel-1, indicating strong capability for capturing deformation gradients and supporting dynamic landslide evolution analysis and early risk identification in geohazard-prone areas.

In conclusion, Fucheng-1 exhibits strong engineering applicability and operational stability for potential landslide monitoring in rugged mountainous terrain. Its high spatial resolution and sensitivity to small-scale deformation make it particularly suitable for geohazard applications. The findings provide an engineering demonstration for the practical deployment of Chinese commercial SAR satellites in geohazard monitoring and offer guidance for sensor selection and future constellation design.