Life Cycle Mining Deformation Monitoring and Analysis Using Sentinel-1 and Radarsat-2 InSAR Time Series

Abstract

The life cycle of mining results in various patterns of surface deformation as it progresses through development, production, and reclamation. Therefore, the spatial–temporal patterns of ground deformation provide a crucial indicator to understand the mining activities, related geohazards, and environmental restoration. This study investigates the decadal deformation (2012–2022) of three coal mines during different stages of mines’ life cycles in Henan, China, using radar interferometry with Radarsat-2 and Sentinel-1 data. The results reveal multiple deformation patterns across different areas: the Changcun mine area changed from ground subsidence to uplift following the termination of exploitation in 2016; the Xiadian mine area has been continuously developing over the past decade, resulting in a cumulative subsidence of 55.6 mm; and the Liyuan mine area exhibits surface rebound at a rate of 7.9 mm/year since its closure in 2007. We also probe the mining geometry of the production process by using a rectangular model. This study highlights the significance of long-term InSAR observations and deformation modeling in elucidating the mining operation dynamics of small mining zones in their production, transition, and post-closure periods, thereby facilitating the management of small-scale mining.

1. Introduction

The closure of small and environmentally unfriendly coal mines is crucial for China’s energy structure transition towards a low-carbon economy. Between 2006 and 2013, over 20,000 small coal mines were closed, with plans to terminate another 15,000 by 2030 in China [1]. However, the process of “mining, closure, and reclamation” in small coal mines (i.e., coal mines with annual production below 0.45 million tons per year) often spans several years to decades. Geological hazards such as surface subsidence, ground fissures, and sinkholes may continue to develop even after the termination of mining [2,3,4]. However, the decadal monitoring of surface deformation in mining areas and goafs remains challenging due to the high labor and economic costs of traditional in situ geodetic observations, impeding subsequent management and ecological restoration efforts in small-scale mining areas [5,6].

In recent decades, multi-temporal InSAR techniques (MT-InSAR), such as Persistent Scatterer InSAR (PS-InSAR) [7,8] and Small Baseline Subset InSAR (SBAS-InSAR) [9,10,11], have become popular for deformation monitoring and modeling in mining areas [12,13,14,15,16,17]. Several studies focus on the surface subsidence after coal mine closures [18,19,20,21,22]. For example, Guéguen et al. (2009) estimated surface displacement 15 years after the closure of coal mines in Nord/Pas-de-Calais, France, using D-InSAR and PS-InSAR. Their findings indicated that the surface deformation gradually stabilized in the post-closure period [23]. Jung et al. (2007) used JERS-1 SAR data from 1992 to 1998 to monitor subsidence in the abandoned Gaeun coal mine in South Korea with PS-InSAR. The results showed an excellent agreement between the InSAR-estimated subsidence and in situ fissure distribution [24]. Zheng et al. (2023) demonstrated a four-stage change in post-closure surface deformation as “subsidence-stabilization-uplift-stabilization” in Xuzhou, China, from 2016–2019 Sentinel-1 data [25]. In Europe, several studies also reported that the groundwater level can change the surface deformation from InSAR observations. For example, Bejarano Urrego et al. (2016) showed that the abandoned Zolder mine in Belgium experienced an uplift of 5.6 mm/year, which was related to the rising groundwater [26]. Gee et al. (2017) reported a linear relationship between the surface uplift and groundwater level from the InSAR observations in Northumberland and Durham coal fields, UK [27]. Blachowski et al. (2018) indicated that a 12 mm/year ground uplift in Walbrzych Coal Basin, Poland may be caused by the increasing hydrostatic water pressure in the rock matrix and the corresponding stress redistributions [28].

These InSAR-based deformation measurements can also be used to probe the geometry and process of mining activities. He et al. (2022) used the multi-source Okada model, a dislocation model for shear and tensile fault motions in an elastic half-space [29], to simulate ground deformation obtained using the coherent scatterer InSAR technique. The results showed that the predicted and observed deformations were highly correlated, with an overall misfit of less than 5 cm [30]. Lu et al. (2013) modeled a mining area with vertical deformation by ignoring the deformation in strike-slip and dip-slip directions and assumed a horizontal tunnel in the Okada model for simplification. Their results showed a high degree of consistency between the leveling and model predictions [31]. Dai et al. (2020) used both the Okada and Mogi models to invert the D-InSAR-based surface deformation by the probability integral method in Huainan coalfield. They found that the more complex and adaptable Okada model performs better at a higher noise level [32]. Ren et al. (2021) also effectively monitored and reversed the deformation of the Dianping coal mine by combining the SBAS-InSAR technique with the dislocation [33].

However, most of the existing studies only capture and model the mining deformation in a short period (i.e., 3–5 years), which cannot cover the entire mining life cycle. Therefore, our understanding of the surface deformation patterns and mechanisms before and after coal mining cessation is still limited. In this study, we first combined Radarsat-2 and Sentinel-1 to achieve continuous monitoring from January 2012 to December 2022 in three mining areas in Henan, China. Second, we employed the rectangular dislocation model and Bayesian inversion method to probe the geometry parameters of the coal seam. Third, we discussed the deformation mechanisms during different stages of the mining life cycle.

2. Materials and Methods

2.1. Study Area

The study area is located in Luoyang and Pingdingshan, Henan, China. This area encompasses two tectonic units: the North China Craton and the Qinling Trough, which have experienced multiple stages of geological activity, leading to the formation of rocks. It hosts abundant mineral resources, including coal, iron, lead, and molybdenum, and has hosted mining operations for several decades. The study area has three primary mining regions (Figure 1a): (1) Longmen mine area has Longmen, Zhengxingsu, and Changcun coal mines (Figure 1b). The Longmen coal mine has been operated since January 2003, producing 0.5 million tons per year across an area of 5.7 km². The Changcun coal mine operated from July 2010 to September 2016, and it covers 18.9 km², with an annual production of 1.8 million tons. The Zhengxingsu coal mine operated from March 2011 to December 2021, with an annual coal production of 0.15 million tons over 1.08 km². (2) Xiadian coal mine is located in Pingdingshan City. The mine’s capacity amounts to 195 million tons in an area of 26.4 km². It has been in continuous operation since May 2010, with an annual coal production capacity of 1.5 million tons. (3) Liyuan coal mine is 17.2 km south of the Xiadian coal mine. The operation lasted for 17 years from December 1990 to September 2007, with an annual production of 0.45 million tons. And the mining style here is longwall mining [34].

2.2. InSAR Data Processing

This study utilized two types of SAR data (

Table 1. SAR data used in this study.

Sensor	Wavelength	Azimuth/Range Pixel Spacing	Orbit Direction	Path	Temporal Coverage
Sentinel-1	5.6 cm	13.94 m/2.33 m	Ascending	113	13 November 2015–30 December 2022
Radarsat-2	5.6 cm	5.22 m/11.84 m	Descending	21571L	31 January 2012–10 January 2016

): (1) Radarsat-2 wide-mode data (11 images) acquired from 31 January 2012 to 10 January 2016; and (2) Sentinel-1A/B TOPS-mode data (190 images) acquired between 13 November 2015, and 30 December 2022, to monitor surface deformation throughout the entire life cycle of the mining areas.

We performed PS-InSAR data processing by differential interferometry, PS point selection, and time series estimation (Figure 2). We first selected the primary images to decrease the potential decorrelation by minimizing the temporal and spatial baselines. The 31 July 2014 and 8 January 2019 were selected to form interferogram stacks of Radarsat-2 and Sentinel-1, respectively (Figure 3a). The interferometric processing included registration, topography, and flat-earth phase removal, and interferogram generation was performed using the GAMMA software. We utilized the 1 arc-sec Shuttle Radar Topography Mission (SRTM) DEM (https://earthexplorer.usgs.gov, accessed on 5 April 2024) to remove topography and flat-earth phases. Precise orbits (https://qc.sentinel1.eo.esa.int, accessed on 5 April 2024) were used for orbital corrections to minimize the noise, and adaptive filtering was applied.Second, we selected the PS points by a two-step method to identify the reliable scatterers for the deformation estimation. We employed the amplitude dispersion thresholds $D_A=0.42$ and $D_A=0.6$ for SBAS for PS point and conducted slowly decorrelating filtered phase (SDFP) point detection. Subsequently, PS candidates were refined by the temporal coherence coefficient ${\gamma }_k$ and the variation in phase residual. Here, the temporal coherence coefficient ${\gamma }_k$ is defined as follows [8]:

$${\gamma }_k=\frac{1}{N}\left|{\sum }_{i=1}^N\exp \left\{j\left({\varphi }_{\mathrm{diff} }^i-{\tilde{\varphi }}^i-{\varphi }_{\theta ,}^{i,u}\right)\right\}\right|$$

(1)

where ${\varphi }_{\mathrm{diff} }^i$ is the differential phase; ${\tilde{\varphi }}^i$ is the spatially correlated phase; and ${\varphi }_{\theta ,}^{i,u}$ is the phase of the spatially uncorrelated look angle error. Then, the three-dimensional phase-unwrapping algorithm was applied to obtain the unwrapped phase ${\phi }_{\mathrm{u}\mathrm{n}\mathrm{w},x,i}$ for each high coherence point:

$${\phi }_{\mathrm{d}\mathrm{e},x,i}\approx {\phi }_{\mathrm{u}\mathrm{n}\mathrm{w},x,i}+{\widehat{\phi }}_{\mathrm{a}\mathrm{t}\mathrm{m},x}^m-{\widehat{\phi }}_{\mathrm{a}\mathrm{t}\mathrm{m},x,i}^s+\mathsf{\Delta }{\widehat{\phi }}_{\mathrm{o}\mathrm{r}\mathrm{b},x}^m-\mathsf{\Delta }{\widehat{\phi }}_{\mathrm{o}\mathrm{r}\mathrm{b},x,i}^s-\mathsf{\Delta }{\phi }_{\theta ,x,i}^{\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{r}}-\mathsf{\Delta }{\phi }_{n,x,i}-2k_{x,i}\pi$$

(2)

where ${\phi }_{\mathrm{d}\mathrm{e},x,i}$ is the target deformation, ${\widehat{\phi }}_{\mathrm{a}\mathrm{t}\mathrm{m},x}^m$ is the atmospheric screen in the primary image, and ${\widehat{\phi }}_{\mathrm{a}\mathrm{t}\mathrm{m},x,i}^s$ is the atmospheric screen in the seceondary image; $\mathsf{\Delta }{\widehat{\phi }}_{\mathrm{o}\mathrm{r}\mathrm{b},x}^p$ and $\mathsf{\Delta }{\widehat{\phi }}_{\mathrm{o}\mathrm{r}\mathrm{b},x,i}^s$ represent the orbital error phases in the primary and secondary images, respectively; $\mathsf{\Delta }{\phi }_{\theta ,x,i}^{\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{r}}$ denotes the noise associated with the residual spatial uncorrelation $\mathsf{\Delta }{\phi }_{n,x,i}$; and $2k_{x,i}\pi$ is the phase integer ambiguity.

The spatially correlated look angle (SCLA) error in the PS pixels of the interferograms is estimated after the phase unwrapping step. The SCLA error is estimated by applying a high pass filter in time and low pass filter in space to the unwrapped phase values. Finally, the SCLA error is subtracted from the remaining phase to obtain the deformation estimates.

Due to the lack of in situ measurements, this study assessed the reliability of the PS-InSAR results by a cross-validation with SBAS-InSAR [35,36]. For SBAS analysis, the maximum temporal and spatial baseline thresholds were set as 96 days and 150 m, respectively. The SBAS stacks contain 538 interferograms (see Figure 3b).

2.3. Deformation Modeling Based on Rectangular Model

To probe the geometry of coal mine production, we applied the Okada rectangular model and the Bayesian inversion method to the Xiadian area, which has been continuously exploited since January 2012. We first subsampled the InSAR velocities from 2012 to 2022 with the quadtree method, retaining a total of 697 data points, which greatly reduced the computational complexity. As the exploitation map is unknown, we used a horizontal rectangular sill with uniform opening to simulate surface deformation, equivalent to the total mining impact. Specifically, according to the dislocation theory in the half-space, the surface displacement field caused by the opening component can be expressed as follows [29]:

$$\left\{\begin{array}{l}{u}_x=-\frac{U_3}{2\pi }\left[\frac{q^2}{R(R+\eta )}-I_3{\sin }^2\delta \right]\\ u_y=-\frac{U_3}{2\pi }\left[\frac{-\tilde{d}q}{R(R+\xi )}-\sin \delta \left\{\frac{\xi q}{R(R+\eta )}-{\tan }^{-1}\frac{\xi \eta }{qR}\right\}-I_1{\sin }^2\delta \right]\\ u_z=-\frac{U_3}{2\pi }\left[\frac{\tilde{y}q}{R(R+\xi )}+\cos \delta \left\{\frac{\xi q}{R(R+\eta )}-{\tan }^{-1}\frac{\xi \eta }{qR}\right\}-I_1{\sin }^2\delta \right]\end{array}\right.$$

(3)

where $u_x$, $u_y$, and $u_z$ are the surface displacement in Cartesian coordinates caused by the opening component $U_3$; $\xi$ and $\eta$ are the coordinates of the particle on the fault plane; and $I_1$, $I_2$, and $I_3$ are the influence coefficients.

Assuming that the horizontal mining activity, the dip angle, strike-slip component, and dip-slip component can be simplified as $\mathsf{\delta }\approx 0$, $U_1=0$, and $U_2=0$, respectively, the surface deformation (Equation (3)) can be simplified as follows [37,38]:

$$u_z=\frac{U_3}{2\pi }\left[\frac{\tilde{y}q}{R(R+\xi )}+\frac{\xi q}{R(R+\xi )}-{\tan }^{-1}\frac{\xi \eta }{qR}\right]$$

(4)

Second, we utilized the open-source Geodetic Bayesian Inversion Software (GBIS) to apply nonlinear inversion to the mining geometry [39]. It employs Bayesian inference to quantify the optimal parameters, and their uncertainties were estimated using the Markov chain Monte Carlo method (MCMC), which incorporated the Metropolis–Hastings algorithm [40]. Specifically, the probability of the parameters can be determined as follows:

$$p(\mathrm{m}\mid \mathrm{d})=\frac{p(\mathrm{d}\mid \mathrm{m})p\left(\mathrm{m}\right)}{p\left(\mathrm{d}\right)}$$

(5)

where $\mathrm{d}$ is the data vector, $\mathrm{m}$ is the set of model parameters, $p(\mathrm{m}\mid \mathrm{d})$ is the likelihood function of $\mathrm{m}$ given $\mathrm{d}$ based on residuals between the data and the model prediction of the observations, $p\left(\mathrm{m}\right)$ expresses the prior information (in the form of a prior joint PDF) of the model parameters, and the denominator is a normalizing constant that is independent of m.

Therefore, the uncertainty gives an indication of the confidence interval of the inverted parameters. Here, we also constrained the initial intervals of the target parameters, considering the range of the deformation area and the minging depth, as follows: length $\in [0,5000]$ m, width and depth $\in [0,2000]$ m. We performed ${10}^6$ iterations and removed the first ${2\times 10}^4$ iterations as the burn-in period because the starting samples may highly depend on initial parameter settings and may not represent the real posterior PDFs.

3. Results

3.1. Deformation in Longmen Area

The Longmen area, located east of the Longmen Grottoes, contains the Longmen mining area, Zhengxingsu mining area, and Changcun mining area (Figure 1). The InSAR data reveal a consistent away-motion in the satellite LOS direction from 2012 to 2016. The average deformation rates for Longmen, Zhengxingsu, and Changcun were 5 mm/year, 10 mm/year, and 8 mm/year, respectively (Figure 4). Correspondingly, the maximum cumulative deformations were 22.4 mm, 38.5 mm, and 34.1 mm. However, in the following period from 2016 to 2022, these mining areas exhibited different deformation characteristics (Figure 5). The Zhengxingsu mining area continued to subside at 2 mm/year, while the Changcun mining area uplifted at 12 mm/year (Figure 6b,c). Notably, the Longmen mining area underwent a complex deformation transition in July 2014, November 2017, and March 2021 (Figure 6a).

To better describe the deformation evolution in the Longmen mining area, we selected the feature points P1-P3 and delineated the cross-profiles A1-A2, B1-B2, and C1-C2 (Figure 4). The Longmen mine area experienced three transition periods as subsidence at 10 mm/year from January 2012 to June 2014. Subsequently, from July 2014 to October 2017, the area underwent uplift at a rate of 4 mm/year. From November 2017 to February 2022, the mining area transformed to subsidence again, accruing a cumulative subsidence of approximately 22 mm. Subsequent to March 2022, the deformation reverted to uplift. However, there is only a single deformation transition within the Changcun mining area in September 2016, when a rate of 8 mm/year subsidence changed to a 12 mm/year uplift (Figure 5 and Figure 6c). In contrast to the transition in the Longmen and Changcun mines, the Zhengxingsu mine experienced a two-stage subsidence, with a significant deceleration measured from 6 mm/year to 2 mm/year in June 2015 (Figure 5 and Figure 6b).

3.2. Deformation in Xiadian Area

The Xiadian region, located in Xiadian Town, Ruzhou City, consists of two main deformation zones that cover 24.4 km². The InSAR result indicates that the Xiadian mining area has been steadily subsiding, with a maximum cumulative deformation of 10 cm (Figure 7).

Figure 7. LOS deformation rate and cumulative deformation time series in the Xiadian region from January 2012 to December 2022 from PS-InSAR: (a) LOS deformation rate obtained from Radarsat-2 data during 2012–2016; (b) time series of cumulative deformation from 2012 to 2016; (c) LOS deformation rate from Sentinel-1 during 2015–2022; and (d) time series of cumulative deformation from 2015 to 2022; D1-D2, E1-E2, and G1-G2 are the section lines drawn in Figure 8 (please note the different ranges of color bars in subplots for visualization).

Figure 7. LOS deformation rate and cumulative deformation time series in the Xiadian region from January 2012 to December 2022 from PS-InSAR: (a) LOS deformation rate obtained from Radarsat-2 data during 2012–2016; (b) time series of cumulative deformation from 2012 to 2016; (c) LOS deformation rate from Sentinel-1 during 2015–2022; and (d) time series of cumulative deformation from 2015 to 2022; D1-D2, E1-E2, and G1-G2 are the section lines drawn in Figure 8 (please note the different ranges of color bars in subplots for visualization).

Two notable deformation areas are Xiadian Town (Point P4) and Qianhu Town (Point P5) in the Xiadian area. In Xiadian town, the ground indicates a non-linear deformation, decreasing from 10 mm/year to 5 mm/year after 2016. However, the Qianhu area shows a continuous subsidence at 10 mm/year, resulting in a 10 cm cumulative deformation and the formation of the deformation profiles (D1-D2, E1-E2, G1-G2). It can be observed that the deformation area in the Xiadian mining area is extensive, with continuous subsidence over 10 years. The subsidence area covers approximately 25.86 km², and from 2016 to 2022, the cumulative deformation reached 50 mm (Figure 9). This resulted in a significant subsidence funnel with a width of approximately 3.1 km.

3.3. Deformation in Liyuan Area

Situated in the southwest of Ruzhou City, close to Jiliao Town, the Liyuan mining region covers an area of about 7 km². The mining area of Liyuan showed stable uplift deformation, with a maximum cumulative deformation of around 70 mm for the observation period of January 2012 to January 2016. Following that, a positive uplift deformation of between 5 and 10 mm/year was observed in the Liyuan mining area between November 2015 and December 2022 (Figure 10).

Figure 10. LOS deformation rate and cumulative deformation time series in the Liyuan region from January 2012 to December 2022 from PS-InSAR: (a) LOS deformation rate obtained from Radarsat-2 data during 2012–2016; (b) time series of cumulative deformation from 2012 to 2016; (c) LOS deformation rate from Sentinel-1 during 2015–2022; and (d) time series of cumulative deformation from 2015 to 2022; F1-F2 is the section line drawn in Figure 11 (please note the different ranges of color bars in subplots for visualization).

Figure 10. LOS deformation rate and cumulative deformation time series in the Liyuan region from January 2012 to December 2022 from PS-InSAR: (a) LOS deformation rate obtained from Radarsat-2 data during 2012–2016; (b) time series of cumulative deformation from 2012 to 2016; (c) LOS deformation rate from Sentinel-1 during 2015–2022; and (d) time series of cumulative deformation from 2015 to 2022; F1-F2 is the section line drawn in Figure 11 (please note the different ranges of color bars in subplots for visualization).

There is only one notable deformation area, which is Jiliao Town (Point P6) in the Liyuan area. The total uplift has reached 7 cm, with an average uplift rate of 5 mm/year. It is obvious that the uplift in the Liyuan mining area has a nonlinear characteristic, and the deformation has stabilized since December 2022. Specifically, from 2012 to 2013, the deformation rate was rather gradual. After 2014, it had a period of significant uplift, which continued until the deformation rate began to stabilize around 2020 (Figure 12).

3.4. Cross-Validation

To validate the reliability of the PS-InSAR results, we employed a cross-validation approach using PS and SBAS techniques in three mining areas (Figure 13, Figure 14 and Figure 15). In Liyuan, Xiadian, and Longmen mining areas, the mean differences in deformation rates are 0.09 mm/year, 0.57 mm/year, and 0.65 mm/year, respectively. The standard deviations of these three areas are 0.98 mm/year, 1.5 mm/year, and 1.2 mm/year, with correlation coefficients of 0.82, 0.45, and 0.72, respectively. These good agreements from both methods confirms the reliability and accuracy of the derived deformation.

3.5. Modeling Parameters

We used the nonlinear Bayesian method to invert the mining geometry in the Xiadian area. The posterior PDFs of inverted parameters reveal that the sill geometry is well constrained after the ${10}^6$ iterations (Figure 16). The overall distribution range of the residual results was [−5 cm, 5 cm], the mean residual value was 0.65 cm, and the standard deviation of the residuals was 1.79 cm, indicating that the surface simulation deformation results based on the Okada model had a high accuracy, and the inverted coal seam mining parameters were reliable (Figure 17c,d). The maximum posterior probability solution and 95% confidence intervals indicate the length, width, and depth of the mining tunnel and are 2634 ± 500 m, 309 ± 98 m, and 711 ± 290 m, respectively (

Table 2. Inverted parameters of mining geometry in Xiadian.

Model Parameters	Optimal Solution
Sill length (m)	${2634.43}_{-531.89}^{+336.40}$
Sill width (m)	${309.01}_{-98.00}^{+97.70}$
Sill depth (m)	${711.08}_{-110.74}^{+292.68}$
Sill strike (deg)	${253.99}_{-20.62}^{+18.62}$
Sill X (m)	${-167.54}_{-215.25}^{+188.57}$
Sill Y (m)	${277.96}_{-98.94}^{+96.94}$
Sill opening (m)	${-1.16}_{-0.58}^{+0.60}$

). It has a SW-NE strike (254 ± 20°).

4. Discussion

This study investigated five mining areas using radar interferometry to comprehensively understand the life cycle stages of mining activities. The InSAR results revealed that the deformation patterns in continuously operated mining areas may undergo multiple transitions (e.g., Longmen mining area) or exhibit subsidence at varying rates (e.g., Zhengxingsu and Xiadian mining areas). After the closure, mine areas generally experience surface uplift, characterized by a linear or exponential function (e.g., Liyuan mining area).

4.1. The Possible Mechanisms for Deformation Transition in Mine Areas

We observed the transition of deformation in the Changcun and Longmen areas. First, in Changcun mine area, the single transition from subsidence to uplift on 14 September 2016 may be related to groundwater. As the groundwater was no longer extracted after the production ceased, the rising level of groundwater can directly result in the surface uplift. This phenomenon has also been observed in other closed mines, such as Limburg mine area in Germany [41]. Recently, Declercq et al. (2023) used Envisat data to map surface rebounds after the closure of the last active coal mine in Zolder, Belgium [42]. They demonstrated that the uplift transition in the western part of the mine can be attributed to the replenish of aquifer after the closure of mine. Similarly, Samsonov et al. (2013) found that after the cessation of the last exploitation at La Houve in Lorraine (French side) at the end of 2004 and the commencement of flooding in June 2006, the deformation reversed from subsidence to uplift [43]. Therefore, although the groundwater records are unavailable in Changcun mine area, according to the temporal development of the surface deformation, it is supposed that the process of groundwater recharge after the mine closure may cause the surface uplift. Complex spatiotemporal deformation characteristics also reveal different responses to surface deformation during different mining life cycles. In the Longmen mine, we observed two rounds of subsidence and uplift. The periods of subsidence from January 2012 to June 2014 and from November 2017 to February 2022 were mainly attributed to phased production activities [44]. Similar to the Changcun mine area, two phases of uplift in the Longmen mine area (i.e., July 2014–October 2017 and March 2022 to December 2022) may be related to the change in groundwater.

4.2. The Deformation Mechanism in Continuously Exploited Mines

The Xiadian mining area has been developed since 2010 and remained in continuous operation during our study period. The deformation is predominantly visible as linear subsidence. There are many factors that lead to mining area subsidence. The compaction of the goaf will drive the subsidence [45], and underground coal fires will also cause ground subsidence [46]. In our study, continuous mining caused a void beneath the surface and, finally, bending and subsiding of the overlying strata. The inverted parameters of the mining tunnel show a rectangular shape with a length of 2634 m, a width of 309 m, and a depth of 711 m.

4.3. The Deformation Mechanism in the Post-Closure Period

In contrast, the Liyuan mining area exhibits a significant difference to the Xiadian mining area. Throughout the 10-year monitoring period, the Liyuan mining area remained in a post-closure state. Following closure, the rising groundwater levels and changes in rock mass structure contributed to an uplift tendency. The uplifting velocity decays from 5 mm/year to 1 mm/year correspond to the exponential tendency of groundwater recovery in the post-closure areas [47].

5. Conclusions

This study utilized Sentinel-1 and Radarsat-2 interferometry to investigate long-term surface deformation across three mining areas in Luoyang and Pingdingshan from 2012 to 2022. We identified the deformation transitions in the Changcun and Longmen areas, which may be related to the rising groundwater after the mine closure and the levels of annual production. For the terminated Liyuan area, an equilibrium was reached, and the continuous uplift ceased in 2021. We also analyzed the mining tunnel in the Xiadian area, indicating a geometry with a depth of 711 m, width of 309 m, and length of 2634 m, with an opening of-1.2 m in the last decade. This study demonstrates the capability of long-term InSAR observations, especially for mining areas without available ground measurements, in deciphering the mine activities and facilitating small mining management throughout a mine’s life cycle. In future, we suggest examining the impact of groundwater on mining activity by coupling hydro-mechanical models, well log data, and stratigraphic information.