Remote Monitoring of Ground Deformation in an Active Landslide Area, Upper Mapocho River Basin, Central Chile, Using DInSAR Technique with PAZ and Sentinel-1 Imagery

Abstract

The upper Mapocho River basin, located in central Chile, has been affected by numerous landslides in the past, which may become more frequent due to a projected increase in intense precipitation events in the context of climate change. Against this background, this study aimed to analyze the ground deformation associated with an active landslide area in the Yerba Loca basin using the SBAS–DInSAR technique with PAZ and Sentinel-1 images acquired during two time periods, 2019–2021 and 2018–2022, respectively. Using PAZ imagery, the estimated vertical displacement velocity (subsidence) was as high as 9.6 mm/year between 2019 and 2021 in the area affected by the Yerba Loca multirotational slide in August 2018. Analysis of Sentinel-1 images indicated a vertical displacement velocity reaching −94 mm/year between 2018 and 2022 in the Yerba Loca landslide, suggesting continued activity in this area. It, therefore, may collapse again soon, affecting tourism services and the local ecosystem. By focusing on a mountainous region, this study demonstrates the usefulness of radar imagery for investigating landslides in remote or hard-to-reach areas, such as the mountain sector of central Chile.

1. Introduction

Landslide processes represent the most frequent geological hazard in mountainous environments, including the Andes mountain range, significantly impacting infrastructure, natural resources, and people’s lives [1,2,3,4,5], with rainfall being the common triggering factor for these processes [6]. Highly intense rainfall events will likely increase in the context of climate change, especially in mountainous areas such as the central Andes mountains of Chile, where a significant increase in the frequency and intensity of rainfall is anticipated [7].

The frequency of these events in the mountainous area of the Metropolitan Region of Chile has an increasing projection in the future according to the variable-resolution modeling implemented in the Community Earth System Model using the RCP 8.5 scenario (representative concentration pathways), with events of up to 260 mm of rain in 24 h [8]. This region has been affected by different types of landslides in recent years [9,10]. Additionally, in the central Andes region of Chile and Argentina (32°S–34°S), prehistoric landslides corresponding to large-volume megaslides (>106 m³) have also been inventoried [10], which indicates that this area is geomorphologically active concerning this type of process.

In this context, one of the main applications of synthetic aperture radar (SAR) remote sensing is the monitoring of areas affected by land surface deformation using the differential interferometric synthetic aperture radar (DInSAR) technique [11,12,13,14,15,16,17]. This technique exploits information contained in the phase of at least two complex SAR images obtained at different times and from different orbital positions over the same area, forming an interferometric pair [12,18].

One tool used for monitoring the temporal evolution of ground deformation is the small baseline subset (SBAS) technique [19], which is a DInSAR algorithm. It is based on an appropriate combination of differential interferograms produced from data pairs characterized by a small orbital separation (baseline) to limit spatial decorrelation phenomena [20,21,22]. Furthermore, a key feature of this algorithm is its ability to reduce phase noise and filter atmospheric artifacts. Thus, the SBAS method is a robust algorithm that provides results regarding both large-scale deformation phenomena and more localized displacement effects [23], enabling the generation of displacement time series maps from a set of SAR data [16].

Various authors have used the DInSAR technique and Sentinel-1 C-band SAR images from satellites of the European Space Agency (ESA) in the Copernicus program framework to monitor landslide processes. In Chile, the authors of [24] analyzed the slide that occurred in 2018 in Yerba Loca, where the accumulated line-of-sight (LOS) displacement between July 2018 and April 2019 was up to 6 cm, with average velocities exceeding 3 cm/year, suggesting that the deformation continues. In [25], results indicated that the landslide remained active between March 2019 and 2021 with speeds above 14 cm/year and a secondary deformation toward the northeast. In [26], it was reported that between 2018 and 2022, subsidence of up to 28 cm occurred, with an average vertical displacement rate of 9.4 cm/year, indicating that the landslide is still active.

In the Precordillera of San Juan, Argentina, the authors of [27] monitored three active areas of mass removal using the DInSAR technique between October 2014 and December 2018 with S1 images. They reported that the average LOS deformation rate for the ascending orbit time series was between −0.76 and 2.29 cm/year, while for the descending orbit series, it was between −2.97 and −0.6 cm/year.

The authors of [2] analyzed surface displacement in the Blanca and Negra mountain ranges, near Carhuaz in Peru, detecting active landslide processes with the DInSAR technique and multiple SAR sensors between 1992 and 2017. The analysis results with S1 images between 2015 and 2017 indicated that the average movement rate was 0.46 mm/year for ascending S1 images and −0.69 mm/year for descending images.

In [28], the authors identified several active landslide sectors using the PSI and SBAS techniques with S1 images between 2015 and 2016 in the city of La Paz, Bolivia, detecting a surface movement rate of up to 158 mm/year to the west and 49 mm/year to the east of the study area.

In Spain, using images from the Spanish PAZ satellite operating in the X band, the authors of [5] analyzed an active slow-moving landslide in an urban area in Alcoy (Alicante) between September 2019 and February 2021. This landslide affected urban infrastructure and transportation in the study area. We observed a displacement rate of up to −31.08 mm/year in the satellite’s line of sight, with accumulated displacements of up to −50.3 mm. Other studies using PAZ and the DInSAR technique have focused on the identifying changes in groundwater levels due to agricultural activity in Turkey [29]; mapping the vertical and horizontal deformation of airport infrastructure in Hong Kong [30]; studying deformation in architectural heritage areas in Portugal [31]; evaluating land subsidence associated with anthropogenic and geological conditions in the eastern region of Recife, Brazil [32]; monitoring hydraulic infrastructure to analyze ground deformation using the European Ground Motion Service and multitemporal InSAR techniques with Sentinel-1, TerraSAR-X, and PAZ images between 2014 and 2021 [33]; and mapping ground movements through SAR co-polarimetric data from PAZ images [34].

Monitoring ground movement in this sector of the central Chilean Andes is essential, as it enables continuous observation of ground deformation and the early detection of potential landslides, particularly in a nearby nature sanctuary that attracts thousands of visitors annually. Furthermore, its proximity to Santiago, Chile’s capital and most populous city, highlights the critical importance of sustained and precise monitoring to mitigate potential risks to the population and the surrounding environment.

This research aimed to monitor the ground deformation associated with an active area of landslides using images from the Spanish PAZ radar and Sentinel-1 images. The analysis used the DInSAR technique to study two time periods (2018–2022 and 2019–2021) and compared the results from the two sensors.

The organization of this article is as follows: After the introduction, background, and definition of objectives, we present the study area and the image collections used. Then, the DInSAR processing algorithm, SBAS, is described. The results shown first correspond to the PAZ images; afterwards, the Sentinel-1 results. This report ends with a discussion and conclusions.

2. Materials and Methods

2.1. Study Area

The study area corresponds to a mountainous sector of the Metropolitan Region of Chile in the Yerba Loca Nature Sanctuary, located at 33°15′15″ (south latitude) and 70°17′14″ (west longitude) (Figure 1). It has a Mediterranean climate [35], with an average rainfall of 335 mm in the past 10 years (Río San Francisco Station formerly Junta Estero Yerba Loca; Chilean Hydrological Agency—DGA), concentrated in winter, and with a prolonged dry season of 7 to 8 months. The annual snowfall recorded at Los Bronces Station was 735 cm in 2023. The relief of the study area is mountainous, with altitudes ranging from 1352 to 5403 m a.s.l., and 53% of the surface of the study area is above 3500 m a.s.l. In addition, 18% of the slopes are above 40°, and the exposure of the slopes is predominantly W- and NW-oriented (42% of the surface of the study area).

Various landslide processes have affected the Yerba Loca basin in recent years. In January 2012, due to heavy rainfall, 15 landslides (debris flows and rock falls) affected the study area [36]. In August 2018, a multirotational slide occurred in the upper part of the Yerba Loca basin, 1210 m above the main watercourse and 7.7 km from Villa Paulina (UTM coordinates: 381.940E and 6319.270N; zone 19, south) [37]. The estimation of the total displaced volume of the landslide was between 2,520,000 m³ and 2,850,000 m³. It occurred weeks after a series of winter precipitation events that generated heavy snowfall, followed by rapid melting conditions [24].

In January 2023, the Casa de Lata sector of the Yerba Loca Nature Sanctuary was affected by a small-volume debris flow estimated to be between ca. 10,000 and ca. 13,500 m³. Additionally, in the upper part of the Ortiga basin, in the Los Nogales Nature Sanctuary (neighboring basin), three rotational slides and two debris flows were recorded, one occurring in January 2023 [38].

Most past landslides in the Yerba Loca basin have occurred in areas where gently to strongly deformed volcanic and volcaniclastic rocks of the Miocene Farellones Formation predominate (Figure 2), locally affected by hydrothermal alteration; other landslides have occurred in glacial, colluvial, and within older landslide deposits of Pleistocene–Holocene age [39,40,41].

Figure 3a shows the hillshade of the Yerba Loca basin, revealing the rugged relief of the study area and demonstrating the geometric distortions, such as shadows and foreshortening, affecting these sectors. Figure 3b shows the ALOS PALSAR digital elevation model in meters above sea level. Figure 3c shows the intensity data from the PAZ image (ascending orbit), and Figure 3d shows the intensity data from the S1 image (ascending orbit). The west-facing slopes exhibit higher backscatter due to antenna illumination.

2.2. Data Collection

To monitor the active sectors of landslide processes in the Yerba Loca basin, two datasets were used from two different periods: images from the Spanish PAZ radar satellite between 2019 and 2021 and Sentinel-1 images from the European Space Agency (ESA) between 2018 and 2022. The analysis periods for the two image sets differed because the National Institute of Aerospace Technology (INTA) of Spain provided the PAZ images under announcement of opportunity AO-001-050 in 2019, which limited data collection to the 2019–2021 period. Figure 1 illustrates the coverage of the PAZ and Sentinel-1 images, and

Table 1. Characteristics of the PAZ and S1 images used in this study.

Main Characteristics of the Images	PAZ Images	S1 Images
Wavelength	X-Band; λ 3.1 cm (9.65 GHz)	C band; λ 5.6 cm (5.405 GHz)
Acquisition mode	Stripmap (SM)	Interferometric wide swath
Processing level	Single-look slant range complex (SSC)	Single look complex (SLC)
Incidence near-angle	~33°	~33°
Incidence far-angle	~36°	~51°
Acquisition dates of the ascending orbit	17 September 2019, and 24 April 2021	21 September 2018, and 22 March 2022
Acquisition dates of the descending orbit	17 September 2019, and 5 June 2021	14 September 2018, and 27 March 2022
Pixel spacing	1.7 × 1.9 m (in range and azimuth)	2.3 × 14 m (in range and azimuth)
Polarization	VV vertical transmission–vertical receipt	VV vertical transmission–vertical receipt
Total number of ascending orbit images	19	131
Total number of descending orbit images	22	194

presents the characteristics of both sensors.

The connection graphs (Figure 4) of both ascending and descending orbit PAZ images show the connection between the super-master image (orange dot) and the slave images (green dots). The black lines represent the interferograms that met the minimum SBAS requirements, and the red dot indicates the image that did not meet the minimum requirement. The mean perpendicular baseline was 139 m and the maximum temporal baseline was 176 days for the ascending orbit. Seventy-seven pairs were analyzed in all combinations in the connection graph in Figure 4a. The super-master image corresponds to the image of 22 November 2019.

The mean perpendicular baseline was 144 m and the maximum temporal baseline was 176 days for the descending orbit. One hundred and four pairs were analyzed in all combinations in the connection graph in Figure 4b. The super-master image corresponded to 11 December 2020.

The connection graphs (Figure 5) of both ascending and descending orbit S1 images show the connection between the super-master image and the slave images. The mean perpendicular baseline was 51 m and the maximum temporal baseline was 120 days for the ascending orbit, and the connection graph in Figure 5a shows that 658 pairs were analyzed in all combinations. The super-master image corresponded to the date of 27 November 2020.

The mean perpendicular baseline was 51.7 m and the maximum temporal baseline was 108 days for the descending orbit, and the connection graphs in Figure 5b show that 977 pairs were analyzed in all combinations. The super-master image corresponded to 22 August 2019.

Combining both satellites leveraged the advantages of PAZ’s high spatial resolution and Sentinel-1’s high temporal coverage and atmospheric penetration capabilities, enabling robust and comprehensive characterization of ground deformation in the study area.

2.3. DInSAR–SBAS Processing

The study monitored the temporal evolution of ground deformation in active landslide areas using the small baseline subset (SBAS) technique [19] and differential SAR interferometry (DInSAR), implemented through the ENVI SARscape 5.7 software [42]. The analysis applied the DInSAR technique to generate interferograms, representing the phase difference between two images acquired at different times over the same area. Equation (1) illustrates the variation in interferometric phase [12,23,43,44,45].

∆φ = ∆ φdis + ∆ φtopo + ∆ φorb + ∆ φatm + ∆ φscatt + ∆ φnoise

(1)

Equation (1). Interferometric phase variation.

The variation in the interferometric phase ∆φ results from ∆ φdis corresponds to the change in distance between the satellite and the Earth’s surface along the direction of the wave pulse (line-of-sight, LOS); ∆ φtopo corresponds to the topographic phase due to non-perfect knowledge of the actual height profile; ∆ φorb corresponds to the orbital phase and explains the residual fringes caused by the use of inaccurate orbital information in the synthesis of the topographic phase; ∆ φatm corresponds to the phase contribution due to the variation in microwave propagation conditions (atmospheric conditions including ionospheric and tropospheric delays) between master and slave image acquisitions; ∆ φscatt corresponds to the phase components due to changes in backscatter behavior; and ∆ φnoise includes all phase noise contributions. The component of the interferometric phase associated with the displacement was eliminated from the contribution of the topographic phase ∆ φtopo using a reference digital elevation model [46].

Figure 6 shows the workflow of the SBAS interferometric process and displacement decomposition in SARscape in 7 steps. First, once we imported the PAZ and Sentinel-1 images into the SARscape format (1), the SAR pair combination and the connection network were defined (2) for the generation of multiple differential interferograms with the temporal and spatial baseline values specified [42]. For both the ascending and descending orbits of the PAZ and S1 images, the maximum normal baseline (%) was used at 45%, corresponding to the maximum percentage of the critical baseline value considered acceptable in generating possible connections. Likewise, the maximum temporal baseline parameter used was 180 days, corresponding to the threshold of the maximum temporal distance between acquisitions, which is considered acceptable. Subsequently, the interferometric process (3) executes automatically the following processing sequence: (a) interferogram generation and flattening, (b) adaptive filter and coherence generation, and (c) phase unwrapping. For (a), we co-registered the images using the SRTM1 DEM and generated differential interferograms based on the connections established in the previous step. For the PAZ and Sentinel-1 images, a multilooking technique was applied with 9 and 4 looks in the range direction, and 7 and 1 looks in the azimuth direction, respectively, resulting in an approximate pixel size of 15 m.

To improve the quality of the interferograms, the Goldstein filter [47] was used (b), and for phase unwrapping (c), the SNAPHU minimum-cost flow (MCF) algorithm was used [48]. In the phase unwrapping, we used a coherence threshold of 0.2 for the PAZ images and 0.3 for the S1 images, so we eliminated all pixels with values lower than this threshold from the analysis. In addition, the precise orbits of the S1 images, which we downloaded from the Copernicus server, were used for processing.

Next, the first model inversion is applied (4) to derive the residual topographic height and the displacement velocity. We carried out the second model inversion (5) to estimate the displacement time series, which we filtered to remove atmospheric phase components to fit the final displacement velocity model. The final stage of the interferometric process (6) involves geocoding the time series data into geographic coordinates, ensuring that all pixels in the image are correctly positioned [42].

Finally, for the PAZ and S1 images, the vertical (up–down) and east–west (E–W) components of the displacement were calculated using the LOS measurements of the ascending and descending geometry using the displacement decomposition tool implemented in SARscape.

3. Results

The following section presents the results of the accumulated line-of-sight (LOS) displacement derived from PAZ images (2019–2021) and Sentinel-1 images (2018–2022), considering both ascending and descending orbits. In addition, we show the decomposition of the displacement into vertical (up–down) and east–west (E–W) components for both image sets within the study area, to analyze ground deformation associated with the active landslide zone.

3.1. LOS Displacement and Components of Movement Vertical (Up–Down) and E–W with PAZ Imagery (2019–2021)

The following section presents the SBAS processing results for the PAZ images. Figure 7 shows a cumulative displacement of the Yerba Loca slide of −4.9 mm in the ascending pass (point a) and 25.8 mm in the descending pass (point c) between 2019 and 2021. Mean deformation velocity LOS displacement was −1.5 mm/year and 14.8 mm/year in ascending and descending orbits, respectively. For the secondary deformation associated with the Yerba Loca rotational slide, a cumulative displacement was 11 mm in the ascending orbit (point b) and 22.8 mm in the descending orbit (point d). The mean deformation velocity LOS displacement was 8 mm/year and 12.5 mm/year in the ascending and descending orbits, respectively. Points a and c in Figure 7 are located within the active Yerba Loca slide, while points b and d are in the secondary landslide. Due to differences in orbital configuration (ascending and descending), the points are not in the same position. However, we have selected them as closely as possible to enable consistent comparison between datasets.

Figure 8 shows the line-of-sight (LOS) displacement time series for points a, b, c, and d, as indicated in Figure 7.

Figure 9 shows seven areas with active ground deformation identified using SBAS–DInSAR to process PAZ imagery. This area correlates with areas with active landslides, close to the Yerba Loca rotational slide that occurred in August 2018, and where a debris flow occurred in January 2023 (La Lata sector).

Figure 9a shows the accumulated vertical component of the movement (mm) in the seven sectors where we identified deformation using PAZ images. Figure 9b shows the E–W displacement in the same sectors, and Figure 9c and 9d show the vertical and E–W displacement velocities (mm/yr), respectively.

The graphs in Figure 10a,b show the vertical and E–W components of the movement detected between 2019 and 2021. In the vertical displacement component, points 2, 3, and 5 show movement acceleration since May 2020. The E–W component shows that points 2, 3, and 5 were moving toward the slope, so these sectors simultaneously show subsidence and displacement in the direction of the slope, which we can observe in the field photographs in Figure 11.

3.2. LOS Displacement and Components of Vertical Movement (Up–Down) and E–W (2012–2022) with S1 Imagery

The following section reports the SBAS processing results for the S1 images. Figure 12 shows the cumulative displacement of the Yerba Loca landslide to be 130 mm in the ascending pass and −509.5 mm in the descending pass between 2018 and 2022. There was a mean deformation velocity LOS displacement of 37.7 mm/year and −164 mm/year in the ascending and descending passes, respectively. For the secondary deformation associated with the Yerba Loca rotational slide, there was a cumulative displacement of −146.8 mm in the ascending pass and −68.2 mm in the descending pass, with a mean deformation velocity LOS displacement of −45.4 mm/year and −19.9 mm/year in the ascending and descending passes, respectively. Figure 12a shows LOS cumulative displacements in the ascending orbit and Figure 12b shows LOS cumulative displacements in the descending orbit. Due to differences in orbital configuration (ascending and descending), the points are not in the same position. However, the selection placed them as close as possible to enable consistent dataset comparison.

Figure 13 shows the line-of-sight (LOS) displacement time series for points a, b, c, and d, as indicated in Figure 12.

Figure 14 shows seven sectors with active deformation, which the DInSAR processing of Sentinel-1 images also identified, some of which (points 1, 2, 3, 5, and 6) coincide with the sectors detected by PAZ. Some points do not coincide because the coverage of the PAZ images is not the same as that of the S1 images. Furthermore, the period analyzed with S1 is more extended (2018–2022), and the number of images used was greater.

Figure 14a shows the cumulative vertical displacement component (mm) in the seven sectors identified with the S1 images. Figure 14b shows the E–W displacement in these same sectors, and Figure 14c and Figure 14d show the vertical and E–W displacement rates (mm/year), respectively.

The graphs in Figure 15 show the vertical and E–W displacement components between 2018 and 2022. In the vertical displacement component, point 6 corresponds to the area where the maximum acceleration in the movement was detected, coinciding with the slide sector in 2018, and showing evidence of continued displacement. Points 2 and 9 reveal acceleration in the movement that began in March 2020. The E–W movement component shows that all points moved in the direction of the slope except for point 2, which corresponds to the secondary deformation. Hence, these sectors show subsidence and ground displacement in the direction of the slope, which the field photographs in Figure 11 also illustrate. Point 2 also depicts displacement in the E–W direction.

Points 1 and 2 correspond in the PAZ and S1 images to a secondary deformation associated with the rotational slide [24] in 2018. It has a lobular shape and a concentric vertical deformation pattern. The vertical movement component shows subsidence of up to −2.9 and −21.7 mm at point 1 and −17 and −42 mm at point 2 according to PAZ and S1, respectively. The length from the scarp to the base of the lobe is approximately 50 m in the deformation associated with point 2, and the displacement velocity is −10 and −43 mm/year according to PAZ and S1, respectively.

Point 3 corresponds to a gelifluction slope at approximately 4000 m a.s.l., with an average gradient of 25°. Gelifluction slopes may be susceptible to triggering landslide processes, so the vertical component of the displacement could indicate that this is a precursor to a landslide.

Point 3 shows maximum subsidence values of −11.7 mm and −46.4 mm for PAZ and S1, respectively. The E–W displacement component reaches up to −25.9 mm and −79.1 mm, indicating that the slope is shifting westward. This sector’s maximum vertical movement velocity is −6.2 mm/year and −15.9 mm/year for PAZ and S1, respectively.

Points 4 and 5 also lie in gelifluction slopes. The sector near point 5 has a maximum vertical movement component of −8.7 mm and −63.7 mm, with a velocity of −4.9 mm/year and −19.3 mm/year for PAZ and S1, respectively.

Point 6 is located where the 2018 Yerba Loca landslide occurred. In this area, the subsidence value reaches a maximum displacement of −288.8 mm, with a maximum velocity of −94.4 mm/year between 2018 and 2022.

Point 7 only provides deformation data based on the PAZ images due to image coverage limitations. This sector shows a lobe shape with high scarps, suggesting that it may have partially collapsed in the past and could also be an area where a new mass movement may develop. The main scarp is approximately 25 m high, while the frontal scarp is 30 m high, ending in a debris flow. This sector is moving downslope (average displacement of −20.5 mm in the E–W direction) and shows material accumulation, with a vertical movement component of up to 16.0 mm.

The S1 images only provide data for points 8 and 9. Point 8 is a rock outcrop close to the 2018 Yerba Loca rotational slide. The vertical displacement component reaches nearly −228 mm, with a maximum velocity of −68.5 mm/year. The E–W movement component indicates that this sector is shifting downslope by −385.8 mm, with a velocity of −117.5 mm/year.

According to the authors of [49], in their hazard assessment using velocity measurements from the DInSAR technique, points 2, 6, 8, and 9, analyzed with S1 images, have a moderate hazard level (between 20 and 100 mm/year, classified as active). In contrast, the remaining sectors present a low hazard level (<20 mm/year). No high-hazard-level areas have been detected so far in the analyzed period. Figure 16 shows Planet images before and after the slide in Yerba Loca, which makes it possible to demonstrate the movement of the scarp in the area.

4. Discussion

The results obtained using the SBAS–DInSAR technique with the time series of PAZ (2019–2021) and S1 (2018–2022) images show that the study area underwent active deformation during the analyzed periods. These results agree with the geomorphological evidence observed in the field and the analyses carried out using SAR images, which also show that the Yerba Loca sector presents active geodynamic processes such as recent landslides, with the last large event occurring in 2018 [24].

The accumulated line-of-sight (LOS) displacement results obtained from PAZ and Sentinel-1 imagery indicate that the Yerba Loca rotational slide remains active. The change in sign of the LOS-projected displacement between descending and ascending orbits, as described in [27], indicates gravitational processes.

The analysis using PAZ images estimates up to −16.5 mm of subsidence in the secondary landslide sector associated with the 2018 rotational slide of Yerba Loca (point 2) and an average velocity of −9.6 mm/year. Concerning the analysis with the 2018–2022 S1 images, subsidence of up to −288 mm is observed, with an average velocity of 94 mm/year in the main 2018 Yerba Loca landslide area.

The results show active deformation between 2018 and 2022, mainly associated with the steep Andean region’s upper valley and gully heads. The deformation was concentrated primarily in places where previous Late Holocene landslides occurred or in areas covered by glacial and periglacial deposits [41]. Thus, it is likely that deformation and, therefore, the initiation of landslide areas in this region concentrates in glacial and previous landslide deposits where unstable permafrost may be present and subject to temperature fluctuations due to climate change, adding more instability to this high and steep mountain region.

Previous results reported in [24,25,50] point out that the pre-Andean sector of the Metropolitan Region presents active geodynamic processes and ground deformation associated with mass removal events. The authors of [24] analyzed the slide that occurred in 2018 in Yerba Loca, considering three deformation periods: (i) the first, between January and July 2018, with a cumulative LOS displacement of 30–70 mm and displacement velocity with an average of 100 mm/year and a maximum of 168 mm/year before the main movement; (ii) the second period, between July 2018 and April 2019, during which the cumulative LOS displacement was up to 60 mm, with average velocities greater than 30 mm/year, suggesting that the deformation continued; (iii) the third period, between May and August 2019, which showed that the deformation continued in both the primary and secondary landslides, with lower values compared to those in previous periods, ranging from 20 to 40 mm/year.

The deformation values estimated in this study with PAZ images were lower than those indicated in [24,25]. However, due to the loss of coherence in the images of the sector, we did not obtain information for a significant part of the Yerba Loca slide. In contrast, the vertical displacement velocity detected by S1 between 2018 and 2022 was similar to the one indicated by [25] between 2019 and 2021 in the 2018 Yerba Loca landslide, and both studies also detected the secondary landslide in the sector. It is important to note that the results obtained in this research do not correspond to absolute measurements, but instead refer to points in the SAR images considered stable or without movement [51] and to the study period analyzed using the PAZ and S1 images.

The DInSAR analysis periods for the PAZ and S1 images did not coincide because we obtained the PAZ images through an opportunity announcement, and the image acquisition was limited to that specific period (2019–2022). However, the PAZ and S1 images suggest active ground deformation in the 2018 Yerba Loca landslide sector. As suggested by [52], this could indicate that these areas remain unstable and deforming, and could be potential areas initiating landslide processes.

The DInSAR technique is important for monitoring active geodynamic processes in mountainous and inaccessible areas such as the Andes of central Chile, as it provides high-coverage images [53]. However, this technique also has some limitations, as the analyzed sectors are located on steep slopes and are affected by geometric distortions such as shadows and foreshortening due to the terrain. Additionally, we may observe low-coherence areas due to temporal decorrelation caused by vegetation changes, snow presence, or atmospheric conditions when the images were captured [54,55]. In this regard, the study area corresponds to a mountainous sector in central Chile at altitudes above 3000 m above sea level, which is a meteorologically unstable area where winter images may show snow coverage. Snow cover would similarly affect the radar signal’s interaction with the surface and the backscatter on different image acquisition dates. Although we did not perform atmospheric correction in the SBAS processing, such as implementing GACOS or ERA5 in the SARscape software, when processing time series with the SBAS technique, the atmospheric phase components were removed to prevent them from affecting the displacement velocity model [42].

The signal penetration is different for each band and is directly related to terrain characteristics such as scatterer distribution, terrain roughness, and the complex dielectric constant [46,51,55,56], in which the C band (S1 images; λ 5.6 cm) has greater penetration than the X band (PAZ images; λ 3.1 cm). In the case of this study, there was more information loss in the PAZ images than there was in the S1 images; however, the period analyzed with S1 (2018–2022) was longer, and the number of images used was also higher compared to that in the PAZ image set. Because of all these factors, different coherence thresholds were used during processing.

5. Conclusions

Finally, this type of research can contribute to developing future strategies for the prevention and mitigation of geohazards in mountainous areas. Considering that the Yerba Loca Nature Sanctuary is a popular destination in the Andes within the Metropolitan Region of Chile, and due to its proximity to the densely populated capital, Santiago, characterized as a recreational area for mountain sports (trekking) and wildlife watching, the occurrence of future landslides in the area may affect the ecosystem and biodiversity of the valley, as well as touristic and recreational activities, posing a significant risk to park visitors. Therefore, SAR image monitoring is an essential tool for detecting the initial formation of such events in remote areas at an early stage [49,52,57,58].

The precise identification of areas with active ground deformation facilitates the delineation of high-risk zones, enabling local authorities (such as the Municipality of Lo Barnechea) to implement timely measures, including real-time intensive monitoring, installing early warning systems, and developing specific evacuation protocols. Additionally, this information can be integrated into territorial planning instruments at both the municipal and regional levels, supporting decisions related to infrastructure development, land-use regulations, and sustainable tourism management. These actions can contribute to minimizing risk while assuring public safety and safeguarding the environmental integrity of the area.