DeformationMonitoring inanAlpineMiningArea in theTianshan Mountains Based on SBAS-InSAR Technology

State Key Laboratory of Frozen Soil Engineering, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou 730000, China Da Xing’anling Observation and Research Station of Frozen-Ground Engineering and Environment, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Jiagedaqi 165000, China College of Resources and Environment, University of Chinese Academy of Sciences, Beijing 100049, China School of Civil and Hydraulic Engineering, Ningxia University, Yinchuan 750021, China


Introduction
Mining activity can cause surface deformations and negatively impacts the environment [1,2]. If rehabilitation measures are not handled properly, they can lead to serious subsequent disasters such as mine collapse [3], slope instability [4], vegetation death over large areas [2,5], and severe water-soil erosion [3]. ese potential threats and problems are more prominent in high-altitude areas due to the fragile habitats therein [2]. e key to solving these problems is to provide theoretical support for decisionmaking based on a large amount of basic data such as surface deformation data. Ground settlement [6] data provide an understanding of changes in mining areas and the surrounding environment, contributing to their sustainable development. Subsidence data can also provide synchronous information on how to develop in areas containing perilous rocks, providing a basic foundation for mining safety in cold-plateau regions.
In recent years, with the development of computer technology, remote sensing (RS), and geographic information system (GIS) technology [7][8][9], increasingly more novel geographic analysis methods [10][11][12] have been used for the dynamic monitoring of surface subsidence in mining areas, among which monitoring based on interferometric synthetic aperture radar (InSAR) technology [13][14][15] has performed excellently. One of the advantages of InSAR over traditional time-consuming and costly monitoring methods, such as level measurements and global positioning satellite (GPS) measurements, is that it has all-day, all-weather [13,16] observation capability. Moreover, microwaves can effectively penetrate the atmosphere and are less affected by water vapor and surface vegetation [17]. In addition, as more and more synthetic aperture radar (SAR) satellites have been launched, increasingly more materials are accessible [16,[18][19][20][21], and the data acquisition cycles have become shorter. As a result, the accuracy of these measurements has greatly improved. InSAR has the added advantage of obtaining ground deformation information over large areas at a relatively small cost. As a corollary, deformation measurements based on InSAR have becoming increasingly popular [21].
For the above reasons, to understand the dynamic changes of subsidence in alpine mining areas, this study focuses on a mining area in the Tianshan Mountains of China as the research object, which consists of 89 scenes of Sentinel-1A level-1 single look complex (SLC) data processed via SBAS-InSAR technology to obtain the surface subsidence [35] information of the area. e subsidence information is analyzed to understand surface subsidence around the mining site and to provide recommendations for mining activities. e cumulative deformation information is also compared with data obtained from in situ global navigation satellite system (GNSS) measurements to verify the accuracy of SBAS-InSAR technology for surface deformation monitoring and subsidence measurements in alpine mining areas. From this, the extent of impact of mining activity on local environment can be understood. In addition, the obtained deformation over a large area can be used as fundamental information to provide a theoretical basic for mining policy, which is of great value to ensure environmental sustainability.  Figure 1(a)). Its altitude ranges from 3,160 m to 4,365 m above the sea level (m.a.s.l.), with an average altitude of 3,839 m. e annual temperature has a maximum of 16°C and a minimum of −30°C [2], and it is a typical high-cold and high-altitude area. e snowline is 3,700-3,900 m.a.s.l., and the areas above the snowline are covered with glaciers, snow cover, permafrost, rock glaciers, and other periglacial geomorphology. e watershed located in the southwest of the mining area divides the rivers into two basins flowing south and north [9]. e south-flowing rivers converge with the Yili River, which is dominated by small rivers. e northflowing rivers empty into Noor Lake, which eventually feeds into the Manas River. e maximum daily precipitation is 146 mm, the annual rainfall exceeds 1,000 mm, the evaporation is 425 mm, the maximum wind speed is 12 m/s, the average humidity is about 43%, and the wind direction is mainly north-northeast [2]. From October to April, the average temperature is below freezing, and solid precipitation such as hail and snowfall dominates this period. In contrast, from July to September, there is relatively little snowfall and warmer temperatures, making this the season appropriate for grass growth. e mine is located along the watershed of the study area, which consists of a large open-pit iron ore mine where the extracted ore is sorted by a mobile crushing and dry sorting station [64]. e geology consists of an erosiondenudation-tectonic alpine landform and a denudationdeposition alpine-valley landform with extensive granite distribution. Two major sections, a tailings stockpile (irregular blue frame in Figure 1(c)) and an active area (irregular black frame in Figure 1(c)), are the focus of this study. e tailings stockpile includes a beneficiation region, and the active area consists of an open-pit mining area (irregular red frame in Figure 1(c)), early prospecting area, and part of the tailings waste stockpile. e rocks around the mining site are heavily weathered [64] and often have fallen rock accumulations. In addition, the rocks are subject to strong cryogenesis, gelifraction, and tectogenesis, inducing a high risk of potential geological hazards. is, coupled with the impact of the mining activities, increases the susceptibility to geological hazards, while frequent weather changes, such as high winds, snow, and hail, can easily cause meteorological disasters and lead to a range of environmental and geological hazard problems.

Data and Methods
Sentinel-1A level-1 SLC products are available for free download from the ASF website (https://vertex.daac.asf. alaska.edu/). e time span is from 25 th December 2017 to 2 nd January 2020, with an average of more than two images per month, excluding cases where no data are available, for a total of 89 scene images. e precise orbital data corresponding to each scene image is available from the ESA website (https://qc.sentinel1.eo.esa.int/). ere are two types of orbital data: one is the precise data created 21 days after the GNSS downlink date, and the other is the restituted product generated within 3 h of receiving the GNSS data. e precision orbital data with ephemerides have positions with accuracies better than 5 cm, and the other data have accuracies of 10 cm. e orbital data used here are the products with precise ephemerides. e digital elevation model (DEM) of the study area is the USGS EROS Archive-Digital Elevation-Shuttle Radar Topography Mission (SRTM) 1 Arc-Second Global, which has a spatial resolution of 30 m and can be obtained from the USGS EarthExplorer website (https://earthexplorer.usgs.gov/). e GNSS static monitoring network system is located at the southwestern foot of the mountain near the mining area ( Figure 1(c)), which can be used to monitor threedimensional (3D) ground surface deformation information, and it consists of nine parts ( Figure 2). e monitoring system which consists of six stations in total, one base station and five observation stations, started working on 9 th October 2019. After the complex calculation between these two kinds of data, the deformation, with a temporal resolution of seconds, can be obtained. However, due to the high temperature variability in the area and the low winter temperatures, part of the stations has experienced some anomalies, and the deformation data monitored at certain times are missing. e GNSS measured deformation data used for the comparison span from 16 th October 2019 to 31 st May 2020 and condition with no data or abnormal data in this period have been rejected.

Study area burst_IW3_1_slc
Mining area Tailings area Activity area To better analyze the settlement characteristics around the mining site, six observation points were selected around each of the tailings stockpile area and the active area containing the mining activities, with six more based on topography, ground cover, and runoff distribution, for a total of 12 points, as shown in Figure 1(c). More information can be seen in Table 1.
e workflow mainly consists of data preprocessing, SBAS-InSAR processing, and data analysis, as shown in Figure 3.

Data Preprocessing.
is process aims to convert the format of the data so that the different software platforms can recognize them and clip the data, thus reducing the time spent on data processing and improving the overall efficiency. Firstly, the region of interest (ROI) is mapped and converted to .shp format. en, the DEM in .tiff format is converted to binary format using SARscape [7], and Global Mapper is applied to create isoheights of the total study area with a spacing distance of 200 m. After these two steps, the SLC data are imported and set at a resolution of 20 m × 20 m. en, the processing images data (raster) area is sectioned with the ROI vector data (.shp). e resolution of the image depends on the pixel spacing slant range, pixel spacing azimuth, incidence angle, and multilook number, which is calculated as where R sr is the range slant resolution, R a is the azimuth resolution, and P sr and P a are the pixel spacings of the slant range and azimuth, respectively, all in units of m. N sr is the number of slant range looks and N a is the number of azimuth looks; both are positive integers. θ is the angle of incidence in degrees. e images used here were obtained from an ascending attitude at an angle of incidence of 39.08°, with pixel spacings of 3.73 m and 13.89 m for the slant range and azimuth, respectively. To ensure that we yielded a target deformation map with a spatial resolution of 20 m × 20 m, the multilook numbers of the slant range and the azimuth, which can be calculated by (1) and (2), were set as five and one, respectively.
Based on the SARscape software for baseline estimation and SBAS data processing, a temporal baseline threshold of 180 d and a spatial baseline threshold of 2% of the critical baseline were set for all data processing to ensure the accuracy of the results and to avoid spatiotemporal decoherence of the data as much as possible.

SBAS-InSAR Processing.
A differential interferogram and coherence intensity graph are obtained by differential interference processing. e final output of this step is a flattened interferogram, where if the constant phase caused by the acquisition geometry and an input DEM are provided, the topographic phase is removed. A Goldstein adaptive filter [65] is chosen to reduce interference from noise phases while outputting the filtered product. Interferometric coherence, which is an indicator of the phase quality and the master intensity filtered image, is also generated. en, the minimum cost flow method [66] with a coherence threshold of 0.35 to unwrap the phase, which is obtained after the above processing, is employed. Finally, the interferometric image pairs are edited according to the unwrapping result, the coherence intensity graph, and the flattened interferogram to ensure that the connection diagram is correct, and image pairs are removed if the value of the coherence intensity graph is low. After the editing process, ground control points (GCP) are selected for refinement and reflattening processing to accurately estimate the orbital parameters so that the unwrapped results are refined and the residual phase is removed. e deformation rate and residual topography are estimated by the first inversion, and the atmospheric phase error is removed by the second inversion. Multiple time series displacements are calculated and the SBAS-InSAR results are geocoded. e results consist of LOS cumulative displacements, in units of mm, and multitime series rate deformation, in units of mm/a.

Data Analysis.
To visualize the displacement results and extract the features of the deformation information, we divided the results into three-year parts with a time interval of about three months in every year. We note that this choice is in keeping with the changing seasons. en, the subsidence rate map was used to analyze the subsidence features around the mining site, select and visualize typical sites, extract their multitime subsidence information to understand the subsidence characteristics, and map the data. Finally, the subsidence information for the comparison points (CPs) was extracted and compared with the GNSS data, which were obtained from observations made during the same measurement period to verify the accuracy of the SBAS-InSAR measurements in the study area. ese two kinds of data were obtained during the same observing period and from the same observing places so that the accumulated sedimentation obtained from the two methods  Advances in Materials Science and Engineering 5 could be compared. e accumulated sedimentation can be calculated as follows: where ΔD is the accumulated sedimentation and D 1 and D 2 are the sedimentations of the observational starting data and end data, respectively, all units in mm.

Baseline Estimation.
e baseline information (Figures 4  and 5) of the Sentinel-1A data was obtained after calculation. e image from 21 st October 2018 (yellow dots in Figures 4  and 5) was finally identified as the super master image, and the remaining images were paired with this master image for interferometric processing, which resulted in 1029 pairs of interferometric images ( Figure 5). e maximum absolute normal baseline was 116.56 m, the minimum absolute baseline was 0.71 m, and the average was 50.51 m. e time baseline ranged from 12 d to 180 d. e results of processing indicated that the data were sufficient for use in the SBAS-InSAR process and they contained little decoherence.

Deformation Velocity.
e deformation rate of the entire study area is shown in Figure 6. Data throughout the study area ranged from −44.80 mm/a to 28.04 mm/a, with significant subsidence at the mining area location, with rates in excess of −40 mm/a. Significant subsidence funnels exist in the northwest and southeast corners of the active area. Combined with the field investigations carried out in September 2018, a large accumulation of tailings waste was found in the northwest corner, distributing a river that flows from the mine site to Noel Lake. At the southeast foot, ground ice is developed, and a large number of turn-hole steel pipes left over from earlier prospecting are distributed, with the tops of the pipes ranging from 1 to 2 m from the present ground surface. e subsidence may be caused by the scouring and transporting actions of the river in the northwest. Besides, sinking in the southeast section of the mining area may be due to melting of this ground ice caused by early prospecting activities. In addition, there is a tendency for accumulation along the eastern part of the active area near the mountain, with an accumulation rate of greater than 20 mm/a, mainly due to rolling accumulation towards the foot of the mountain as the dangerous rock is broken by gelifraction, frost action, and physical weathering, while activities in the mine also likely increase accumulation at the foot of the mountain. e deformation rate characteristics of the tailings pile area are similar to those of the active and mining areas, with significant subsidence in the waste piles distributed along the river in the west and uplift in the western part near the hills. Settlement is mostly pronounced in the northwest and southwest, with deformation rates exceeding −35 mm/a in the north and averaging 30 mm/a in the southwest, while in the east, at the foot of the hillside, the piles are clearly deposited at rates greater than 25 mm/a. e relationship between the deformation velocity and altitude ( Figure 7) shows that as the altitude rises, the velocity gets larger. Combined with Table 1 and Figures 1 and  6, we can understand that P1-P6 are located in the tailings area, which is used for mineral dressing and tailings stockpiling, while P1-P3 are located to the west of the area, and P4-P6 are near the mountain to the east of the tailings area. e deformation velocity shows that with the exception of P5, which is located in the accumulation zone, all other points are located in the settlement zone, with P1, P3, and P4 all having settlement rates in excess of 20 mm/a. P7-P12 are included in the activity area, where P8, P9, and P10 are located in the northwest of the area, P7 is in the central part near the mining working area, and P11 and P12 are in the southeast; indeed, all points except P9 are within the mining area.
e deformation velocities of P7, P8, and P11 are positive and the other three are negative.
In general, the area over which the tailings were deposited has settled significantly, with an average settlement rate of over 35 mm/a and more than 40 mm/a in areas that experience river flowing. Mining activity may accelerate the surrounding perilous rock with frost fissure and frost splitting. Due to the gelifraction of glaciers and ice and the engineering blasting of the mining activity, there is large accumulation of sediment at the foot of the slope around the mining site. e settlement rate here is over 20 mm/a, which is a cause of concern. e prolonged and rapid accumulation of debris is a potential risk that could threaten the safety of mining personnel, and as the tailings area and mining area are connected by tunnels (Figure 1(c)), the safety of traffic should be a further concern.

Deformation Characteristics.
Taking the LOS deformation on 25 th December 2017 as a starting point, the deformation of each subsequent scene image was calculated according to (3). e 89-scene deformation results were divided into three groups at three-month intervals, with each group including a whole freeze-thaw cycle in a year, which allowed for better consistency with seasonal changes. e results are shown in Figures 8-10 Figure 6: Deformation velocity in the study area.  From Figures 7 and 11, it can be seen that there is good correspondence between the deformation piling volume and the deformation rate value. e settlement volumes at P1, P3, and P4 increased the most with time, corresponding to the largest deformation rates. e settlement volumes of points P6, P8, and P11 were the second largest, and the settlement rates accelerated after September 2019. e largest amount of accumulation occurred at P5, with an accumulation of more than 70 mm over the three years, which should be of key concern due to its proximity to the tunnel entrance. P9, P10, and P12 showed the same trend of change with uplift, but the rates of uplift were not large and fluctuated with a certain periodicity, which may be due to the regular manual cleaning and removal of the accumulation. P2 and P7 showed no obvious changes and only slight subsidence. rough the fieldwork and site investigations, it was found that P2 is in the location of the processing plant and P7 is the coal mining ladder of the mining area, both of which are relatively stable, so the deformations therein were not obvious.  In summary, the tailings accumulation area and the active area mainly showed subsidence in the west along the river distribution section, while they showed uplift in the east near the foot of the mountain. e reason for this phenomenon is mainly due to the presence of a large number of tailings piles distributed along the river area, which are loose structures that are prone to collapse and can accelerate the subsidence phenomenon due to the transport and scouring effects of runoff.
e area of accumulation is mainly due to the accumulation of mountain debris slides. To ensure the sustainable development and safe operation of the mine, the mine should minimize the accumulation of tailings in the river flow in the future to reduce water pollution, remove the accumulation of tailings, and carry out vegetation restoration. e piles around the mine should be cleaned regularly to ensure safe mining, and the piles at the tunnel entrance and exit locations should be monitored over a large area for a long period of time and dealt with in a timely manner.  Figure 9: LOS deformation of 2019, in mm.

Deformation Precision.
To verify the accuracy of the SBAS-InSAR measurement results, GNSS monitoring data and SBAS-InSAR measurements obtained from 16 th October 2019 to 31 st May 2020 were selected. e cumulative shape variables for the corresponding dates were calculated using (3), and the results were compared. eoretically, the location of a GNSS observation point should be exactly the same as the location of a SBAS-InSAR measurement point; however, due to the reason of decoherence, the SBAS-InSAR monitoring results at the precise GNSS installation locations were missing, so points close to each GNSS installation location were selected as the corresponding CPs to obtain the required data for the observation period. e GNSS data were collected at an interval of 2 h, and the temporal resolution of the SBAS-InSAR monitoring results was 12 days. e results are shown in Figure 12.
As can be seen from Figure 12, the accumulations for each set of corresponding points have a high degree of agreement in the general trend, but do not match perfectly, and the GNSS observations are more prominent than the  Figure 10: LOS deformation of 2020, in mm.

SBAS-InSAR measurements in terms of variability.
is phenomenon is mainly due to the inconsistent temporal resolution of the two measurements, the incomplete agreement of the observation locations, and the inconsistent orientation of the deformation values. e deformations measured by SBAS-InSAR are in the LOS direction, while the GNSS deformation measurements were selected along the x direction. Although there are many limitations that lead to some deviations in the two results, the SBAS-InSAR deformation measurements in this area have high accuracy and credibility. ese results have an important reference value and can provide a theoretical basis for the restoration of mining environments and the creation of safety measures in mining areas.
In summary, the surface deformation measurement results of alpine mining areas based on SBAS-InSAR technology are highly credible and relatively accurate. Although there is a slight difference from the GNSS field observation results, the measurement results of this method lie in its wide observation range, workload, and less human and material resource requirements. GNSS measurements are highly accurate but costly, due to the large surface undulations in alpine areas and inconvenient transportation, coupled with strong frost splitting, frost stirring, Advances in Materials Science and Engineering congeliturbation, and frost jacking effects. Furthermore, it is difficult to keep the GNSS mounting base fixed, which can easily lead to the collapse of GNSS observation instruments, thus affecting the accuracy of the monitoring results; indeed, frequent repair and maintenance will increase the cost of GNSS observations. erefore, SBAS-InSAR surface deformation monitoring based on Sentinel-1A has great prospects for applications in alpine and high-altitude mining areas, its monitoring results can provide a theoretical basis for environmental restoration and safety measures for construction activities in mining areas, and its accuracy is sufficient to meet general surveying and mapping requirements.

Discussion
Surface subsidence measurements based on SBAS-InSAR in alpine and high-altitude areas have proven to provide a wealth of information for mining sites, thus contributing to their sustainable development. However, there are some corresponding problems, such as the deformation information at the location of the mining area not being measured, the presence of a large number of blank areas throughout the study area, and the measured deformation results being in the LOS direction rather than of the vertical rise and fall of mining area deformation. e above problems are mainly due to the decoherence of the data during differential interference. e location of the study mine is a low-lying gully, and the Sentinel-1A data used in this case are ascending, which has resulted in a significant amount of shadowing, layover, and foreshortening [16] due to the topography. In addition, there are a number of limitations inherent [14,16] to InSAR deformation measurements, including various other phase and noise effects, as well as the correct/incorrect setting of parameters during processing. e influence of the phase manifests itself in the atmospheric phase, the noise of the system itself, and in the terrain phase [16,21]. e influence of the parameter settings is mainly in the correct setting of the multiview numbers, the selected filtering methods, and the choice of phase unwrapping methods [66,67]. e contributions of many scholars in recent years have led to an increasing number of parameters [17] to be set, and how to choose a set of parameters suitable for data processing in the study area requires constant trial and error and summation to obtain the optimal values. e decoherence of the data will result in the measured region being partially masked out and thus appearing as a blank in the deformation result map. e results of deformation measurements are values in the LOS direction [14], which are limited by the InSAR measurements themselves. To obtain the real 3D deformation information of the surface, it is necessary to use simultaneous multisource data, multi-incidence angle data, etc., to obtain deformation information along more directions in the study area, so that the true 3D information of the surface can be solved with the mathematical model [4,36,37]. erefore, to obtain accurate 3D surface deformation information, more data must be processed to obtain more deformation information from different aspects, in addition to combining field measurements, retrieving missing measurements by spatial interpolation, etc.

Conclusions
In alpine and high-altitude regions, due to the development of periglacial landforms such as ice field, ground ice, and rock glaciers, open-pit mining involves extensive surface excavation and massive accumulation of surface tailings, which can cause substantial disturbance to the ecological environment and loosen dangerous rocks around the mining site, thus posing a series of potential threats and risks. In this paper, the deformation information and annual average subsidence rate data of a mining site in Tianshan Mountain from 25 th December 2017 to 2 nd January 2021 were obtained by SBAS-InSAR processing of 89-scene Sentinel-1A ascending SLC data, and the measurement results were compared with field GNSS deformation measurements to verify the former's measurement accuracy. e conclusions of this study are as follows: (1) Deformation around the mine is significant compared to the rest of the surrounding area. e maximum sedimentation rate in the study area was 44.8 mm/a, and the maximum uplift was 28.04 mm/ a. e maximum sedimentation rate in the tailings area was 35 mm/a, and the maximum uplift was 25 mm/a. e active area, including the mining area, had a maximum sedimentation rate of over 40 mm/a and an accumulation rate of 20 mm/a. (2) e overall deformation in the study area is dominated by subsidence, with the maximum subsidence amount reaching 129.39 mm during the three years. e uplift phenomenon is weaker than the subsidence phenomenon, with a maximum uplift accumulation of 60.49 mm during the three years, which is about half of the maximum subsidence. e tailings accumulation area and the active area are stacked significantly at the foot of the hill near the mountain area, which showed a three-year uplift of over 75 mm, and the tailings accumulation area distributed along the river in the west settled significantly, with a threeyear settlement of over 80 mm.
(3) e deformation measurement results based on SBAS-InSAR are highly accurate and can provide a theoretical basis for environmental restoration and safety measures for construction in mining areas. Compared with GNSS measurements, the method is less costly and less labor intensive. e measurement results cover a large area, including the entire mining site and the surrounding environment, thus providing a macroscopic understanding of the impact of mining activities on the surrounding environment. e data obtained from this method have the potential for widespread use.

Data Availability
Sentinel-1A level-1 SLC products were downloaded from the ASF website (https://vertex.daac.asf.alaska.edu/); orbital data are available from the ESA website (https://qc.sentinel1. eo.esa.int/); DEM can be obtained from the USGS Earth-Explorer website (https://earthexplorer.usgs.gov/); and other data are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest.