Surface Deformation Monitoring and Risk Mapping in the Surroundings of the Solotvyno Salt Mine (Ukraine) between 1992 and 2021

Abstract

The historical Ukrainian rock-salt mining town of Solotvyno and its environmentally related problems are well-known. A complex monitoring system is needed to evaluate the current situation in order to revitalize the investigated area. In addition to other risks, surface deformation due to undermining is one of the major risks endangering building infrastructure in the inhabited area of the town. These processes are well-known in the area, and damages caused by the surface movement are often recognized. Measurement of the process’s intensity and identification of the impacted area are crucial for any revitalization work. Information on these processes is the most important element of the hazard management and spatial-developmental planning of the town. This study aimed to characterize the long-term surface deformation processes and to identify the spatial and temporal trends and changes of these processes to assist spatial planning. The first step was to understand the surface deformation history from 1992. An InSAR-based assessment of the surface displacement of the undermined Solotvyno area was performed using data from three satellites, namely the ERS, Envisat, and the Sentinel-1, covering the time period between 1992 and 2021. The derived quantitative analysis indicated an intensive surface displacement and subsidence over the mining area. However, these displacements have not been even in the last 30 years of the investigation. The identification of the stabilized areas and recently started movements indicated the dislocation of the processes, which requires adequate actions for geohazard management and strategic planning. The demonstrated technology (InSAR) has the potential to set up an appropriate alarm system and provides an automated mechanism for continuous risk detection. A complex systems development is able to significantly reduce the geohazards over the unstable built-up zones.

1. Introduction and Aims

Mining activity affects most of the European countries where underground mines were operated [1]. The collapse of underground caverns often triggers the subsidence of the covering sediments [2,3]. The subsidence often remains hidden for a longer period of time, and it is perceived only by building damages [4] or rapidly developing devastating surface distortions such as sinkholes [5]. Salt mines, which have hundreds of years of mining history, are in the focus of mining subsidence monitoring [6] due to their intense distortions and the potential to release salted water as pollution during the flooding of their caverns [7]. Therefore, the potential risk emerged by salt mine subsidence and collapses [8] is well-known across Europe [9,10,11]. Surface deformations, subsidence, and displacements are all identified in the region as coinciding and simultaneous processes. These terms will be used in the paper as synonyms, but they are slightly different terms for the acting processes.

One of the most frequently investigated European sites is the abandoned salt mine of Solotvyno in Ukraine. Salt mining was one of the major income sources of the Carpathian basin from the Roman time to the time of the Hungarian kingdom. Salt mines spread from the current Eastern Slovakia (Presov) down to the salt regions of Transylvania (Praid, Turda). Mining operations continued in the area up to 2010. Salt mines and their related traditions are still affecting these regions, acting as touristic attractions and sources of income. However, there are several mining-induced risks in the area that affect the life quality and the environmental and economic sustainability of the area. The Ukrainian Solotvyno is one of the most representative sites of both this tradition and the environment suffering from the post-mining impacts.

The salt mine disaster in Solotvyno has had negative environmental effects in the larger area also. In 2008, the first indications showed significant changes in the water quality of the Tisza River, meaning that the salt content in the river reached the highest sodium chloride contamination value ever detected before; it was ten times the average level. Since then, it was proven that the salt contamination in the Tisza River originated from the collapsed salt mine of Solotvyno. In addition to the salt contamination that pollutes the ground and surface waters, the collapsing mine shafts and openings imply continuous danger to human lives. In 2010, the Ministry of Emergencies of Ukraine announced a state of emergency in the Solotvyno salt mining area.

The Tisza River is the most important cross-border recipient and transmitter factor. Through an overall investigation, the relevant stakeholders and decision-makers will have a starting point for the sustainable revitalization of the Solotvyno mine area. Without this benchmark, any other future interventions are just speculations.

The study will directly help the self-government of Solotvyno and inform the local people on revitalization process. The local monitoring system will connect to and combine with the upper Tisza transnational water quality monitoring system.

The monitoring of mining-induced surface subsidence is costly and often not feasible by onsite measurements [12]. The analysis of remotely sensed imagery archives is one of the most promising tools to map and track historical displacements [13]. Ray et al. coherently reviewed the use of remote sensing techniques for surface movements such as landslide detection [14]. Waldir et al. used integrated SAR analysis techniques to map surface deformation processes in an open pit iron mine in Brazil and concluded that the space-borne SAR data used together with field monitoring is an effective way of surface deformation monitoring [15]. Szűcs et al. [7] demonstrated the effectiveness of the use of an interferometric synthetic aperture radar (InSAR) for surface deformation and subsidence monitoring for the period between 2014 and 2018, using Sentinel-1 data for the Solotvyno area. They concluded that the evolution of the surface changes due to mining activities can be detected. This study aimed to extend the study for a longer period between 1992 and 2021 in order to characterize the long-term processes and forecast the potential changes, their geographic extent, and speed.

Advanced interferometric techniques such as Small Baseline Subsets (SBAS) [16] and Persistent Scatterers (PS) [17] are designed to monitor small-scale surface distortions with millimeter or submillimeter accuracy [18]. These are excellent tools to monitor remote locations that are not accessible, lack preliminarily installed geodetic networks or ground control points (GCP), or are simply too big to perform onsite measurement campaigns. The interferometric monitoring of the surface subsidence of salt mines was the state-of-the-art technique applied in this study [15,19,20].

ERS, Envisat imagery archives, and adjacent Sentinel-1 measurements released by the European Space Agency provide excellent raw material for mapping and monitoring surface deformation using the above-mentioned techniques. The archives contain ca. 30 years of acquisitions of sites that can be utilized to run PS or SBAS algorithms to extract deformation history with submillimeter accuracy in medium resolution (ca. 25 and 15 m). Both processing algorithms were designed for the same purpose; however, since their processing strategies are fundamentally different, each of them can be used to check accuracy and cross-validate the results.

The aims of our study were the following:

• To map the deformations in the vicinity of the Solotvyno Salt Mine between 1992 and 2021 in order to receive accurate spatial and temporal data of surface development;

• To determine the spatial and temporal changes of movement tendencies of the area;

• To prepare a risk map for Solotvyno based on the results of interferometric processing and available auxiliary data sources.

The analysis of the spatial and temporal development of the displacements can provide exact information of the surfaces that were stabilized and of those that are under continuous risk of deformations. Moreover, the spatial and temporal trends of subsidence can also be predicted. This type of information provides critical information for long-term urban planning and development of the town.

2. Materials and Methods

2.1. The Study Area

The study area of the Solotvyno is located in the western part of Ukraine (Figure 1) on the Ukrainian–Romanian border. The town and the mines are located on the older terraces and the recent floodplain of the Tysa river. The highest part of the town rises to 325 m above sea level onto the southern slopes of the Magura mountains, while the lowest part at the Tysa floodplain is a little bit over 250 m elevation. The average elevation is about 280 m.

The town is located in the Marmarosh/Solotvyno basin divided by the Tysa River. The Solotvyno village is situated in the westernmost part of Ukraine in the Tyachev district of the Zakarpattia Oblast (Transcarpathian region) at the Romanian and Ukrainian border (Figure 1). There are two salt mining towns on the two sides of Tysa, Solotvyno on the Ukrainian side and Sighetu Marmației on the Romanian side. The mines are located mainly on the recent floodplain and the older terraces of the Tysa. The salt dome area is located between the Vihorlat-Gutin (Gutaiu) volcanic range and the sedimentary flisch zone of the northeastern Carpathians [21].

The extent of the salt dome under Solotvyno is shown on Figure 2. The village lies on the alluvial plain of the Tisa River, south of the Magura Mountains. The majority of the mining area is located on the Tysa terrace, 20–30 m above the current floodplain, but a smaller area on the western part falls on the low-lying current floodplain (Figure 3). Young sediments of the river were placed over a thick salt deposit originating from the Miocene Carpathian Sea. Younger Holocene clay layers cover the top of the salt dome (anticline structure), which functions as an insulating layer over the underlying salt strata [22]. Mining activity started here during Roman times, and the first subsurface mine was established only in the late 1770s, called Kristina [23], and lasted till 2010. Salt mining was accompanied by water pumping to prevent the flooding of mine shafts during the last century. Due to economic changes, the pumping stations were abandoned, and cavities were flooded first in 1998 [23].

The town has a northwestern and a southeastern flank, divided by a lower-lying meadow area. This area has been developed due to the subsidence caused by the undermining. Its surface is characterized by several small lakes and sinkholes.

The deepest underground tunnels reached a ca. 400–450 m depth from the surface [24]. The first mine collapse happened in 1973 [25] in the central part of the salt dome. Due to the removal of protective layers, incoming water triggered technogenic activation of the salt dome, which turned into significant subsidence and the collapse of the central zone of the dome [26]. The first significant collapses and building damages started in 2000 and were repeated systematically in 2005 and 2008 [23]. Sinkhole formation became a regular process here, reaching a maximum diameter of 200 m. The last mines (mines 8 and 9) were closed in 2010, and the area remained abandoned [27]. The incoming water formed caverns in the salt dome [25], which triggered further sinkhole formation on the surface [28]. Surface subsidence and sinkhole development continuously endanger the settlement; therefore, intense study has been carried out in the last few decades. These investigations were focused on separated time periods of the surface development of the area; however, synthesis of the spatial and temporal development of the subsidence trends over the last 30 years is still lacking.

2.2. Satellite Data

In order to investigate the surface deformations of the last ca. 30 years in the surroundings of Solotvyno, ERS, Envisat, and Sentinel-1 SAR images acquired by the ESA satellites were processed. As a raw material, archive data of ERS and Envisat satellite were downloaded from ESA’s Online Dissemination Portal [29], while Sentinel-1 satellite images were downloaded from the Vertex server of the Alaska Sar Facility [30] as single look complexes (SLC) (Table 1, Figure 4).

2.3. Data-Processing Techniques

Interferometric stacking algorithms such as SBAS and PS are designed to extract displacement of surface scatterers, where phase stability is appropriate for long-term measurements. Multiple images are used to overcome the limitations of basic InSAR and differential DInSAR approaches, with modeling and eliminating atmospheric artifacts and modeling both the topography and displacements. These two techniques focus on different types of objects of a given site. PS focuses on scatterers that have well-defined geometry (e.g., scatterers of urban environment) and good phase stability, while SBAS is designed to analyze the backscattered signal of distributed scatterers (e.g., open fields without well-defined geometry) [18].

All the interferometric analysis of SAR data was done separately according to their sensor and orbits, using the PS and SBAS modules of the Envi 5.6 and SARscape 5.6 software. The SRTM-3(v4) digital elevation model (DEM) was used for the topographic phase removal and for geocoding the results [31].

2.3.1. Interferometric Preprocessing

Data preprocessing consisted of the execution of PS and SBAS algorithms on the mage stacks detailed in Table 1. Both of the algorithms required importing the images and applying the precise orbit files. Afterward, the research site was cut from the images for further processing. Considering the PS algorithm, a connection graph of the images was first created, where one master image was selected from the middle of the image stack and all the other images were considered as slaves. Within the following interferometric step, differential interferograms of the master–slave image pairs were created by also evaluating the digital elevation model (SRTM3 v4) of the site. During the next steps, displacements were modeled, and then the atmospheric phase screen (APS) was modeled and removed. The APS-free modeled displacement data were then geocoded again using the same DEM. The setup of the SBAS algorithm started also with the connection graph creation. Several masters and slaves and one super-master image were defined according to their relative temporal and spatial (baseline length) positions. The processing continued with creating differential interferograms, which were filtered and flattened afterward. Both the displacements and APS were modeled, and the APS was removed. The results were geocoded. SBAS algorithm operates with multiple unwrapping of the interferometric phase and therefore requires more frequent control by the operator, while the PS algorithm is semi-automated.

The displacement data consisted of scatterer points in the case of PS, and raster cells for the SBAS. Both the vector points and raster cells contained the most sufficient descriptors of the given scatterers (annual velocity (mm/y), displacements by date, estimated height, etc.) and also the quality measures (coherence, height precision, and velocity precision). However, it should be noted that all the displacements were compared to the first date of the stack and considered in radar geometry along the line of sight (LOS).

Considering the advanced spatial resolution and therefore the better spatial coverage of scatterers provided by the S1 sensor, both LOS displacements and east–west and vertical displacements were separately calculated and analyzed.

Both the ascending and descending LOS data were the composites of 3D (east–west, north–south, and vertical) displacements [30], which can be expressed as the following (in case of right-looking SAR antenna such as S1):

$\left[\cos \left(\theta \right)-\sin \left(\theta \right)\cos \left(\alpha \right)\sin \left(\theta \right)\sin \left(\alpha \right)\right] \left[\begin{array}{c}{D}_U\\ D_E\\ D_N\end{array}\right]=\left[\mathsf{\Delta }R\right]$

where α is the azimuth of the satellite, θ is the incidence angle at the selected surface point, D_U is the vertical displacement, D_E is the east–west displacement, and D_N is the displacement toward the north. However, calculating the latter (D_N) component was cumbersome, since at least three radar geometries would be necessary and they are not provided by the orbital system of S1. On the contrary, the azimuth angles of S1 are very close to the south–north direction; hence, the system is less sensitive to the displacements toward the north and the south, and the D_N component could be neglected [32]. A system of two equations can be used to describe the relationship of ascending (DLOS_ASC) and descending (DLOS_DESC) LOS measures of a given pixel, where, of course, the east–west and vertical displacement components are the same.

$\{\begin{array}{c}{D}_E\cos {\varphi }_{ASC}\sin {\lambda }_{ASC}+D_U\cos {\lambda }_{ASC}=D_{LOS}{}_{ASC}\\ D_E\cos {\varphi }_{DESC}\sin {\lambda }_{DESC}+D_U\cos {\lambda }_{DESC}=D_{LOS}{}_{DESC}\end{array}$

The D_E and D_U displacement components were calculated using the “Meta Combination” module of Envi SARscape 5.6 in case of SBAS results, while the “Shape Combination” module of the same software was evaluated on PS data for the decomposition [32,33].

As a second aspect of the decomposition of the LOS measurement, it should be noted that the spatial coverage of the ERS and Envisat measurements provided poor overlap of the ascending and descending geometries. Therefore, the decomposition of east–west and vertical components could led to massive data loss, considering these sensors.

To prevent data loss induced by the combination of geometries, only LOS data were kept for further analysis, considering all three sensors. Hence, the inter-sensor combination of displacements focused on LOS data to provide temporal continuity. It should be noted that the displacements had a very strong vertical component and a minor, usually negligible east–west component; hence, the LOS data clearly reflect real displacement tendencies.

2.3.2. Data Fusion and Data Analysis

The processing of the images covered the wide surroundings (30 km) of the research site. Former authors reported both mining subsidence within the surroundings of the abandoned mines and mass movements on the southern slopes of the Magura mountains—not related to the mining activities [25]. Displacements in the vicinity of the salt mine are sporadic and not necessarily connected to the mining subsidence processes; generally, individual structural movements or mass movements are detected at a few places. Therefore, the geographic extent of the data fusion and analysis was reduced. Since the occurrence of mass movements are more frequent on the steeper slopes of the mountains, we calculated the slope map of the site (based on the same SRTM 3 DEM used for interferometric processing). Slopes from the northern edge of the terrace of the Tisa River were excluded from the investigations. On the other hand, surfaces that were within the 1 km buffer zone of the salt-rock boundary (0 m a.s.l.) were kept for further processing (Figure 5).

Due to the different system and acquisition designs of the three SAR satellites, the displacement data differed from each other regarding spatial and temporal resolution. ERS and Envisat satellites had a 25 m spatial and 35 days—or usually more—temporal resolution, while S1 acquisitions were consistent with a 12-later 6-days repeat cycle with 15 m spatial resolution. All of these conditions and the ca. 30 years of study period resulted in significantly changing positions and spatial coverage of the scatterers between decades/sensors. Hence, the biggest challenge was the comparison of displacements measured by different sensors during the data analysis, i.e, generating a consistent input data system.

According to previous research [7,25], field reports [24,27,28], and interferometric measures of our study, all the significant displacements that occurred within the research site consisted of a strong linear subsidence component and negligible horizontal movements. From the SAR sensor point of view, it meant that a given subsiding point of the surface had the same displacements in both ascending and descending geometries. Therefore, measurements of the given sensor calculated in any geometry could be summarized and handled together. To prove the above-mentioned statement, annual velocity values of scatterers of ascending and descending geometries were displayed on scatter plots according to sensors and processing algorithms. Moreover, the correlation coefficients of these ascending and descending annual velocities were calculated (Figure 6).

Figure 4 clearly shows that there was a strong positive correlation in all of the cases, which suggested that ascending and descending geometries can be combined. Therefore, annual velocities of the ascending and descending geometry calculated for the given sensor were merged together, considering both algorithms.

After the merging process, the annual velocity values according to the geometries’ grid networks were created over the study area with 25, 50, and 100 m rectangular cell size to sum the average velocity values of the three sensors separately. The aim of this step was to summarize one velocity value for each cell and to make the velocities measured by each sensor comparable. This methodological step helped to overcome the difficulty caused by the different scatterers of the same location being measured by different sensors. For selecting the optimal grid size, grid velocities were displayed on density plots (Figure 7).

This statistical investigation highlighted that there were no significant differences and data loss among the three cases over the study site. Therefore, a 100 m grid network was kept for further calculations since it was the most convenient case for data visualization and best supported the spatial comparison of surfaces sparsely populated by scatterers (mainly due to Envisat PS and SBAS poor spatial coverage). Hence, annual velocities representing the given sensor were displayed in this 100 m grid network.

For comparing the spatial and temporal appearances of stable and unstable areas, a 1.5 mm/y velocity threshold was applied. This threshold is considered in the literature as an accurate annual velocity estimation by PS and SBAS algorithms on ERS and Envisat images [34,35,36,37,38,39]. On the other hand, S1 measurements were considered more accurate [40]; however, the same threshold was applied to allow inter-sensor comparison. Therefore, a grid moving slower than ±1.5 mm/y was considered stable, and all the others were considered unstable. Stable and unstable grids were later compared between sensors representing the main campaign periods of acquisitions (see Figure 8) to display both the spatial and temporal evolution of the displacements. Based on the stability of the grids, eight categories were formulated (Table 2).

The time series of areas characterized by different movements over time were plotted using their PS time series (considering all the InSAR measurements) and the R software.

3. Results

The main challenge of this investigation was to harmonize the measurements of different sensors (ERS, ENVISAT, and Sentinel-1). As it was outlined in the previous section, a grid system was implemented to provide a general framework for the outputs of different InSAR data processing. In addition to the sensor comparisons, each sensor’s measurements were evaluated individually. Using this approach, both the sensor-specific tendencies over the operation period (Section 3.1 and Section 3.2) and a complex review of the 30 years’ surface deformation trends in the Solotvyno Salt Mine area (Section 3.3) were presented and interpreted.

3.1. Interferometric Results (Annual Velocity of the Site According to Sensors)

Sensor-specific investigations highlight similar spatial trends in the InSAR measurements. In addition to the different sensitivities of the sensors, clear spatial and temporal trends are indicated by InSAR data. Subsidence processes concentrated above the former mining area, and the outer skirt of the salt dome started to stabilize during the last 30 years (Figure 9, Figure 10 and Figure 11). It is important to note that the C-band sensors are insensitive to the vegetated area and water bodies. Therefore, it was hard to collect information over the former mining sites where the sinkholes are located. The figures present the detectable surfaces (grid cells) over the study region according to the sensor sensibility.

The previous figures precisely demonstrate the surface movement velocity of the study area and also point out the variation between detectable areas by sensors. These two aspects must be elaborated separately. The range of the LOS velocity (mm/y) did not change significantly during the decades, but the territory of the intensive movement shrunk the sinkholes. The shrinking is a positive tendency from a geohazard point of view, because the northern part of the actively moving area—endangering the eastern part of the northwestern unit of the village—has been stabilized. It seems quite evident that the Sentinel-1 sensor has the highest sensitivity and coverage over the area. Practically the entire build up area of Solotvyno is observable (Figure 11). It is the main advantage of Sentinel-1 (Figure 11). The areas adjacent to the sinkholes have the highest risk. The mass of Magura hillfoot also can contribute to the displacement processes, especially on the northern edge of the mapped area.

While analyzing the S1-based vertical displacements of the site, we saw the same trends described just above (Figure 12). It soon became obvious that the main displacement components were observable along the vertical axis using both interferometric stacking techniques.

3.2. Stable Movement and Unstable Sites (According to Sensors)

The results of the detailed InSAR processes were transformed into a stability map. A threshold value of 1.5 mm/y annual velocity was set to classify the grid cells into stable and unstable classes. In general, the contiguous blocks of red cells (Figure 13, Figure 14 and Figure 15) indicate the sites with subsidence risk. Some dispersed red cells could be the result of the structural instability of manmade constructions. It must be noted that a very low or sensitive threshold value was used to provide a map of the hazardous zones that may overestimate the intensively moving areas.

Figure 13, Figure 14 and Figure 15 show the differences between the sensors. Similarly to the previous InSAR analysis, the undermined areas are affected the most by the surface movement. The instability of the border region in Figure 15 (Sentinel-1 processing) can be attributed the slope movements and could also be the result of the data processing, due the missing adjacent values of the processing grid.

3.3. Multitemporal Changes of Stable and Unstable Sites and the Temporal Evolution of Unstable Sites

Creating clusters of the grids reflecting temporal differences of stabilization and destabilization helped us to interpret surface evolution of the site. It must be noted that only the grid cells that have been observable in all sensors were presented. Surfaces which show permanent stability lie on the southern and northern edge of the grid network, while the grids which were continuously unstable during the last 30 years (see “permanent instability” on Figure 16) were concentrated in the middle of the site, over the mining area. It is also well-visible that displacements moved back from the sides of the area toward the center of the site during the last 20 years. Therefore, grids which were stabilized only after the 1990s (ERS instability) or after the 2000s (S1 stability) showed zonal patterns between the surfaces defined by permanent stability and instability. This zonal pattern was more remarkable on the southern flank of the site, where the surfaces were stabilized with a greater extent compared to the northern flank of the town. Grids showing interim destabilization (see “Envisat instability”, “Envisat stability”, and “S1 instability” on Figure 16) showed a sporadic pattern without any well-defined spatial correlation. Hence, these grids can be interpreted as the results of displacements caused by individual scatterers within the grid, and their movements were independent from the mining subsidence.

Figure 17, Figure 18, Figure 19, Figure 20, Figure 21, Figure 22 and Figure 23 below show the general displacements measured within the grids according to typical grid clusters of Figure 16. The cluster named “ERS instability” (Figure 17 and Figure 18) is characterized by two representative points, one from the northern flank and one from the southern flank of the village. The two points reflect different moving tendencies. The northern flank moved faster during the 1990s, but both were destabilized.

The grid points of the “permanent stability” cluster (Figure 19 and Figure 20) did not exceed the 1.5 mm/y velocity threshold in any sensors; therefore, there were no well-defined grids oncoming or distancing toward or form the satellite visible on the graphs.

The “S1 stability” cluster is represented by strong LOS displacements of the 1990s and 2000s (Figure 21 and Figure 22) where the cumulative displacements reached 110–130 mm, while these surfaces stabilized in the Sentinel-1 measurement period. Only the grid displayed on Figure 21 reflects periodic displacements, but the sinusoidal movements were likely due to thermal dilatation and were independent from the mining subsidence.

The cluster called “permanent instability” reflects the ongoing mining subsidence (Figure 23). These pixels reached more than 50 cm-es of displacements during the last decades, causing serious threat over the former mining site. The results match with the corresponding literature. Szűcs et al. [7] reported over 15 cm subsidence within 4.5 years for the area, which was slightly lower than our results.

4. Discussion

Several studies have been conducted to test the InSAR techniques to characterize surface deformation processes in mining areas, both for open pit mining [15] and for underground mining [7,25,41,42]. Two recent studies targeted the Solotvyno area [7,25], focusing mainly on the sinkhole formation, forecasting, and development, accompanied by surface deformation processes. Velasco et al. [25] could not identify any stable point around the sinkholes and identified surface deformation speeds up to 10 cm/year. Their study covered only a four-year period, and no long-term characteristics were identified for the region. Szűcs et al. [7] performed their study for the period between 2014 and 2019. Their study showed continuous subsidence with a linear trend. A cumulative line-of-sight deformation of up to nearly 15 cm in 4.5 years was calculated. They found no accelerated or decelerated movement within the 4.5-year period. Their study neglected the central mining area since no stable positions for InSAR monitoring were found in the central—unpopulated grassland and wet areas—part of the town, but the urbanized northern and southern flanks showed a significant surface distortion of up 25 mm/y.

Our study extended the study period from 1991 to 2021 to perform a long-term investigation in order to identify the spatial and temporal trends of acceleration or deceleration of surface movements to support strategic spatial planning. Despite the previous studies that used single-satellite data, this study used the ERS, Envisat, and the Sentinel-1 data combined to achieve a long-term surface movement information. Short periods of investigation would not necessarily provide the detail to identify areas of stabilization, where urbanization can be reintroduced and the area can be revitalized. It was also important to check if there was any shift of the actively moving areas to forecast geohazard risk. This study introduced a procedure and hazard classification system to characterize the surface distortion processes and their temporal trends of stabilization or destabilization within the 29-year period (Figure 16, Figure 17, Figure 18, Figure 19, Figure 20, Figure 21, Figure 22 and Figure 23). The areas characterized by permanent instability matched well with the surface evidences of building damages and sinkhole, lake, and wetland formations due to the significant subsidence. The rest of the classes can be used to identify new risk areas or stabilizing areas where development can be reintroduced.

5. Summary

An InSAR-based assessment of the surface displacement of the undermined Solotvyno area was performed using data from three satellites, namely the ERS, Envisat, and the Sentinel-1, covering the time period between 1992 and 2021. Our quantitative analysis indicated an intensive surface displacement and subsidence over the mining area. However, the extent of the unstable surfaces was reduced and concentrated over center part of Solotvyno. Thirty years of InSAR measurements pointed out that the stabilization processes have developed from outside to the core part of the former mining area. This core region still represents high risk and instability, but the mine-related geohazard has scaled down in the last 30 years in Solotvyno. The area of the unstable surface has halfway decreased in the last 30 years. Despite the positive trends, a considerable surface motion was still detected over the former mining area, exceeding more than half a meter of subsidence, and the trend is not decreasing. Regardless of the favorable processes, the vulnerability of some regions of Solotvyno still exists, and it carries potential risk to the town and its development. This fact implies that the continuous monitoring of the settlement is important, especially the subsidence-prone part, and its surrounding is essential to avoid further damages to the town. The demonstrated technology (InSAR) has the potential to set up an appropriate alarm system and provides an automated mechanism for continuous risk detection. A complex system, together with artificially installed corner reflectors over the center part that is missing observable objects (core area of the mining site), is able to significantly reduce the geohazards over the unstable built-up zones.

Author Contributions

Conceptualization, E.D., L.R. and P.S.; methodology, I.P.K., D.M.K. and L.R.; software, I.P.K., D.M.K. and L.R.; formal analysis, E.D., I.P.K., D.M.K. and L.R.; investigation, E.D., I.P.K., D.M.K. and L.R.; writing—original draft preparation, D.M.K.; writing—review and editing, E.D., I.P.K., D.M.K., P.S., V.M., L.P. and L.R.; visualization, E.D., I.P.K., D.M.K. and L.R.; supervision, E.D., P.S. and V.M. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The study did not require ethical approval.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available on request from the corresponding author. Remotely sensed data are downloadable from the corresponding sites.

Conflicts of Interest

The authors declare no conflict of interest.

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Figures and Tables

Figure 1. The location of the study area, presented at 25 m resolution EU-Dem database version 1.1 (https://land.copernicus.eu/imagery-in-situ/eu-dem/eu-dem-v1.1, accessed on 26 April 2022).

Figure 2. General sketch of the study site. Data is based on Velasco et al. and Shekhunova et al. [24,25]. ©OpenStreetMap contributors.

Figure 3. The delineation of the mining area. Data is based on Velasco et al. and Shekhunova et al. [24,25], presented at 25 m resolution EU-Dem database version 1.1 (https://land.copernicus.eu/imagery-in-situ/eu-dem/eu-dem-v1.1, accessed on 26 April 2022).

Figure 4. Overview of satellite acquisitions used in our study.

Figure 5. Slope map of the research site.

Figure 6. Correlation plots of ascending and descending annual velocities according to sensors and algorithms. In case of the ERS sensor—where only descending data were available—the results of the two algorithms were compared, not the two geometries.

Figure 7. Distribution of S1 average velocity values in the 25, 50, and 100 m grid cells.

Figure 8. Methodological workflow of the technical study.

Figure 9. Average LOS velocity (mm/y) of ERS scatterers (1992–2000).

Figure 10. Average LOS velocity (mm/y) of Envisat scatterers (2002–2010).

Figure 11. Average LOS velocity (mm/y) of S1 scatterers (2014–2021).

Figure 12. S1-based vertical displacements of PS scatterers at the study site.

Figure 13. Stability map of ERS scatterers (1992–2000).

Figure 14. Stability map of Envisat scatterers (2002–2010).

Figure 15. Stability map of S1 scatterers (2014–2021).

Figure 16. Multitemporal stability map of Solotvyno (markers represent plots, see below).

Figure 17. LOS displacements of grids showing ERS instability in the northern flank of the village (ID1 on Figure 16).

Figure 18. LOS displacements of grids showing ERS instability in the southern flank of the village (ID2 on Figure 16).

Figure 19. LOS displacements of grids showing permanent stability in the northern flank of the village (ID3 on Figure 16).

Figure 20. LOS displacements of grids showing permanent stability in the southern flank of the village (ID4 on Figure 16).

Figure 21. LOS displacements of grids showing S1 stability in the northern flank of the village (ID5 on Figure 16).

Figure 22. LOS displacements of grids showing S1 stability in the southern flank of the village (ID6 on Figure 16).

Figure 23. LOS displacements of grids showing permanent instability (ID7 on Figure 16).

Table 1

Images downloaded from the research site.

Name of the Satellite and Geometry	Geometry	Start and End Dates of Acquisitions	Number of Images	Spatial Resolution (in meters)
ERS	descending	May 1992–December 2000	42	25
Envisat	ascending	October 2002–August 2009	25
Envisat	descending	October 2002–October 2010	31
Sentinel-1	ascending	October 2014–October 2021	341	15
Sentinel-1	descending	October 2014–October 2021	337	15

Table 2

Categories of grid stability (⨁ = stability, ⊖ = instability).

Terminology	Definition	Stability and Instability during Acquisitions
		ERS (1990s)	Envisat (2000s)	S1(2014–2021)
permanent stability	The grid was stable over the last 30 years.	⊕	⊕	⊕
permanent instability	The grid was unstable over the last 30 years.	⊖	⊖	⊖
ERS stability	The grid was stable during ERS acquisitions and became unstable.	⊕	⊖	⊖
Envisat stability	The grid was unstable during ERS acquisitions, became stable during Envisat acquisitions, and became unstable again.	⊖	⊕	⊖
S1 stability	The grid was permanently instable during ERS and Envisat but stable during S1 acquisitions.	⊖	⊖	⊕
ERS instability	The grid was instable during ERS but became stable on later acquisitions.	⊖	⊕	⊕
Envisat instability	Instability occurred during Envisat acquisitions.	⊕	⊖	⊕
S1 instability	The grid was permanently stable during ERS and Envisat but destabilized during S1 acquisitions.	⊕	⊕	⊖

References

1. Herrero, C.; Catalina, J.C.; Muñoz, A. Prediction and Monitoring of Subsidence Hazards above Coal Mines (Presidence); European Commission: Brussels, Belgium, 2012; [DOI: https://dx.doi.org/10.2777/3356]

2. Whyatt, J.; Varley, F. Catastrophic failures of underground evaporite mines. Proceedings of the Twenty-Seventh International Conference on Ground Control in Mining; Morgantown, WV, USA, 29–31 July 2008; West Virginia University: Morgantown, WV, USA, 2008; pp. 113-122.

3. Gourdier, S.; Bazargan-Sabet, B. Numerical Model for Assessment of Subsidence Due to Dissolution of Salt, Application to Nancy Basin (East of France). 2014; Available online: https://hal-brgm.archives-ouvertes.fr/hal-01025668 (accessed on 26 April 2022).

4. Diao, X.; Wu, K.; Chen, R.; Yang, J. Identifying the Cause of Abnormal Building Damage in Mining Subsidence Areas using InSAR Technology. IEEE Access; 2019; 7, pp. 172296-172304. [DOI: https://dx.doi.org/10.1109/ACCESS.2019.2956094]

5. Neal, J.T. Mine-Induced Sinkholes over the U.S. Strategic Petroleum Reserve (SPR) Storage Facility at Weeks Island, Louisiana: Geologic Mitigation and Environmental Monitoring. 1997; Available online: https://www.osti.gov/servlets/purl/459946 (accessed on 26 April 2022).

6. Ege, J.R. Formation of Solution-Subsidence Sinkholes above Salt Beds; Geological Survey Circular 897; United States Department of the Interior: Washington, DC, USA, 1984.

7. Szűcs, E.; Gönczy, S.; Bozsó, I.; Bányai, L.; Szakacs, A.; Szárnya Cs Wesztergom, V. Evolution of surface deformation related to salt-extraction-caused sinkholes in Solotvyno (Ukraine) revealed by Sentinel-1 radar interferometry. Nat. Hazards Earth Syst. Sci.; 2021; 21, pp. 977-993. [DOI: https://dx.doi.org/10.5194/nhess-21-977-2021]

8. Delgado, J.M.P.Q.; Guimarães, A.S.; de Freitas, V.P.; Antepara i KoIí, V.; Herný, R. Salt Damage and Rising Damp Treatment in Building Structures. Adv. Mater. Sci. Eng.; 2016; 2016, 1280894. [DOI: https://dx.doi.org/10.1155/2016/1280894]

9. Móga, J.; Gönczi, S.; Berghauer, S.; Móga, K. Georesource or geohazard? Past, present and future of the salt mines in Solotvyno. GeoMetodika; 2019; 3, pp. 5-19. [DOI: https://dx.doi.org/10.26888/GEOMET.2019.3.2.1]

10. Móga, J.; Lippman, L.; Tombor, E.; Fehér, K.; Kéri, A.; Borsodi, A. Geomorphological investigation of the Aknaszlatina saltkarst (Ukraine). Karsztfejlődés; 2015; 20, pp. 185-213.

11. Móga, J.; Szabó, J.; Gönczy, S.; Lippmann, L.; Bódai, B. The study of the dynamically changing landforms of Aknaszlatina saltkarst by field and GIS methods. Karsztfejlődés; 2017; 22, pp. 139-161.

12. Schuchová, K.; Lenart, J. Geomorphology of old and abandoned underground mines: Review and future challenges. Prog. Phys. Geogr.; 2020; 44, pp. 791-813. [DOI: https://dx.doi.org/10.1177/0309133320917314]

13. Mancini, F.; Stecchi, F.; Stecchi, F.; Zanni, M.; Gabbianelli, G. Monitoring ground subsidence induced by salt mining in the city of Tuzla (Bosnia and Herzegovina). Environ. Geol.; 2008; 58, pp. 381-389. [DOI: https://dx.doi.org/10.1007/s00254-008-1597-1]

14. Ray, R.L.; Lazzari, M.; Olutimehin, T. Remote Sensing Approaches and Related Techniques to Map and Study Landslides. Landslides: Investigation and Monitoring; IntechOpen: London, UK, 2020; pp. 1-25. [DOI: https://dx.doi.org/10.5772/intechopen.93681] ISBN 978-1-78985-824-2

15. Paradella, W.R.; Ferretti, A.; Mura, J.C.; Colombo, D.; Gama, F.F.; Tamburini, A.; Santos, A.R.; Novali, F.; Galo, M.; Camargo, P.O. et al. Mapping surface deformation in open pit iron mines of Carajás Province (Amazon Region) using an integrated SAR analysis. Eng. Geol.; 2015; 193, pp. 61-78. [DOI: https://dx.doi.org/10.1016/j.enggeo.2015.04.015]

16. Berardino, P.; Fornaro, G.; Lanari, R.; Santosi, E. A New Algorithm for Surface Deformation Monitoring Based on Small Baseline Differential SAR Interferograms. IEEE Trans. Geosci. Remote Sens.; 2002; 40, pp. 2375-2383. [DOI: https://dx.doi.org/10.1109/TGRS.2002.803792]

17. Ferretti, A.; Prati, C.; Rocca, F. Permanent scatterers in SAR interferometry. IEEE Trans. Geosci. Remote Sens.; 2001; 39, pp. 8-20. [DOI: https://dx.doi.org/10.1109/36.898661]

18. Pasquali, P.; Cantone, A.; Riccardi, P.; Defilippi, M.; Ogushi, F.; Gagliano, S.; Tamura, M. Mapping of ground deformations with interferometric stacking techniques. Land Applications of Radar Remote Sensing; Holecz, F.; Pasquali, P.; Milisavljevic, N. InTechOpen: London, UK, 2014; pp. 231-258.

19. Wasowski, J.; Bovenga, F.; Nutricato, R.; Conte, D.; Refice, A.; Kowalski, Z.; Graniczny, M. Satellite interferometry reveals spatial patterns of subsidence in the ancient Wieliczka salt mine (UNESCO heritage site, Poland). Proceedings of the FRINGE 2007 Workshop; Frascati, Italy, 26–30 November 2007.

20. Liu, X.; Xing, X.; Wen, D.; Chen, L.; Yuan, Z.; Liu, B.; Tan, J. Mining-Induced Time-Series Deformation Investigation Based on SBAS-InSAR Technique: A Case Study of Drilling Water Solution Rock Salt Mine. Sensors; 2019; 19, 5511. [DOI: https://dx.doi.org/10.3390/s19245511] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31847164]

21. Kuzovenko, V.V. Geologic Map of Pre-Quaternary Formations; Transcarpathian Series L-34-XXXV (Uzhhorod), L-34-V (Satu Mare), Scale 1:200,000; Ministry of Ecology and Natural Resources of Ukraine: Kyiv, Ukraine, Department of Geology and Subsurface Use SGE “ZAKHIDUKRGEOLOGIYA”: Kyiv, Ukraine, 2001; (In Russian)

22. Diakiv, V.; Bilonizhka, P. The characteristics of the geological structure and present day geoecological condition of Solotvyno salt bed. Bull. Lviv. Univ. Geol.; 2010; 24, pp. 62-79.

23. Poklad, A. Ecological disaster in Solotvino. Proceedings of the 16th International Scientific Conference Economics for Ecology (ISCS’2010); Sumy, Ukraine, 5–8 May 2010; pp. 120-122.

24. Shekhunova, S.B.; Aleksieienkova, M.V.; Stadnichenko, S.M.; Siumar, N.P. The Integrated Geological Model of Solotvyno Structure as a Tool to Assess Geoecological Sustainability of Solotvyno Rocksalt Deposit. Collect. Sci. Work. Inst. Geol. Sci. NAS Ukr.; 2015; 8, pp. 233-250. [DOI: https://dx.doi.org/10.30836/igs.2522-9753.2015.146791]

25. Velasco, V.; Sanchez, C.; Papoutsis, I.; Antoniadi, S.; Kontoes, C.; Aifantopoulou, D.; Paralykidis, S. Ground deformation mapping and monitoring of salt mines using InSAR technology. Proceedings of the Solution Mining Research Institute Fall 2017 Technical Conference; Münster, Germany, 25–26 September 2017.

26. Onencan, A.M.; Meesters, K.; Van de Walle, B. Methodology for Participatory GIS Risk Mapping and Citizen Science for Solotvyno Salt Mines. Remote Sens.; 2018; 10, 1828. [DOI: https://dx.doi.org/10.3390/rs10111828]

27. Stoeckl, L.; Banks, V.; Shekhunova, S.B.; Yakovlev, Y. The hydrogeological situation after salt-mine collapses at Solotvyno, Ukraine. J. Hydrol. Reg. Stud.; 2020; 30, 100701. [DOI: https://dx.doi.org/10.1016/j.ejrh.2020.100701]

28. Yakovlev, E.O.; Shekhunova, S.B.; Aleksieienkova, M.V.; Siumar, N.P. Assessment of complex stress-strain of the Solotvyno salt anticline structure (basing on technique of natural pulse electromagnetic field of the Earth). Collect. Sci. Work. Inst. Geol. Sci. NAS Ukr.; 2016; 9, pp. 83-96. [DOI: https://dx.doi.org/10.30836/igs.2522-9753.2016.144246]

29. ESA’s Online Dissemination Portal. Available online: https://esar-ds.eo.esa.int/oads/access/collection (accessed on 26 April 2022).

30. Alaska Sar Facility. Available online: https://search.asf.alaska.edu/#/ (accessed on 26 April 2022).

31. Jarvis, A.; Reuter, H.I.; Nelson, A.; Guevara, E. Hole-Filled SRTM for the Globe Version 4. 2008; Available online: http://srtm.csi.cgiar.org (accessed on 26 April 2022).

32. Ng, A.H.M.; Ge, L.; Zhang, K.; Chang, H.C.; Li, X.; Rizos, C.; Omura, M. Deformation mapping in three dimensions for underground mining using InSAR–Southern highland coalfield in New South Wales, Australia. Int. J. Remote Sens.; 2011; 32, pp. 7227-7256. [DOI: https://dx.doi.org/10.1080/01431161.2010.519741]

33. Fialko, Y.; Simons, M. The complete (3-D) surface displacement in the epicentral area of the 1999 Mw 7.1 Hector Mine earthquake, California, from space geodetic observations. Geophys. Res. Lett.; 2001; 28, pp. 3063-3066. [DOI: https://dx.doi.org/10.1029/2001GL013174]

34. Hanssen, R.F. Satellite radar interferometry for deformation monitoring: A priori assessment of feasibility and accuracy. Int. J. Appl. Earth Obs.; 2005; 6, pp. 253-260. [DOI: https://dx.doi.org/10.1016/j.jag.2004.10.004]

35. Colesanti, C.; Wasowski, J. Investigating landslides with space-borne Synthetic Aperture Radar (SAR) interferometry. Eng. Geol.; 2006; 88, pp. 173-199. [DOI: https://dx.doi.org/10.1016/j.enggeo.2006.09.013]

36. Casu, F.; Manzo, M.; Lanari, R. A quantitative assessment of the SBAS algorithm performance for surface deformation retrieval from DInSAR data. Remote Sens. Environ.; 2006; 102, pp. 195-210. [DOI: https://dx.doi.org/10.1016/j.rse.2006.01.023]

37. Crosetto, M.; Monserrat, O.; Iglesias, R.; Crippa, B. Persistent scatterer interferometry: Potential, limits and initial C- and X-band comparison. Photogramm. Eng. Remote Sens.; 2010; 76, pp. 1061-1069. [DOI: https://dx.doi.org/10.14358/PERS.76.9.1061]

38. Cigna, F.; Bianchini, S.; Casagli, N. How to assess landslide activity and intensity with Persistent Scatterer Interferometry (PSI): The PSI-based matrix approach. Landslides; 2013; 10, pp. 267-283. [DOI: https://dx.doi.org/10.1007/s10346-012-0335-7]

39. Pratesi, F.; Tapete, D.; Terenzi, G.; Del Ventisette, C.; Moretti, S. Rating health and stability of engineering structures via classification indexes of InSAR Persistent Scatterers. Int. J. Appl. Earth Obs. Geoinf.; 2015; 40, pp. 81-90. [DOI: https://dx.doi.org/10.1016/j.jag.2015.04.012]

40. Fiaschi, S.; Holohan, E.P.; Sheehy, M.; Floris, M. PS-InSAR Analysis of Sentinel-1 Data for Detecting Ground Motion in Temperate Oceanic Climate Zones: A Case Study in the Republic of Ireland. Remote Sens.; 2019; 11, 348. [DOI: https://dx.doi.org/10.3390/rs11030348]

41. Kim, J.-W.; Lu, Z.; Degrandpre, K.; Kim, J.-W.; Lu, Z.; Degrandpre, K. Ongoing Deformation of Sinkholes in Wink, Texas, Observed by Time-Series Sentinel-1A SAR Interferometry (Preliminary Results). Remote Sens.; 2016; 8, 313. [DOI: https://dx.doi.org/10.3390/rs8040313]

42. La Rosa, A.; Pagli, C.; Molli, G.; Casu, F.; De Luca, C.; Pieroni, A.; D’Amato Avanzi, G. Growth of a sinkhole in a seismic zone of the northern Apennines (Italy). Nat. Hazards Earth Syst. Sci.; 2018; 18, pp. 2355-2366. [DOI: https://dx.doi.org/10.5194/nhess-18-2355-2018]