A 35-Year Analysis of Vegetation Cover in Rare-Earth Mining Areas Using Landsat Data

Abstract

Fractional vegetation cover (FVC) plays a significant role in assessing ecological quality and protection, as well as soil and water conservation. As a typical rare-earth resource county in China, Dingnan County has experienced rapid development due to rare-earth mining, resulting in significant alterations to vegetation cover. To elucidate the spatio-temporal changes in vegetation within Dingnan County over the past 35 years and the effects of natural and human factors on these changes, the spatial and temporal variations in FVC were analyzed using Landsat-TM/OLI multispectral images taken in 1988, 1995, 1997, 2002, 2006, 2013, 2017, and 2023. The findings indicate that (1) vegetation coverage in Dingnan County decreased from 1988 to 2002, followed by a gradual increase; (2) high vegetation cover is predominantly found in forested areas that maintain their natural state, while the central town and mining areas exhibit generally low coverage; (3) there are regional differences in the relationship between vegetation cover and environmental factors in Dingnan County. This research facilitates the alignment of ion-type rare-earth mining with ecological protection, thereby promoting the sustainable development of the mining area and providing scientific guidance for local governments to formulate more effective management and protection strategies for the mining ecosystem. Additionally, this research offers a scientific foundation for mining areas globally to develop sustainable policies and informed decision-making regarding environmental protection and sustainable development.

1. Introduction

As an important and sensitive ecological parameter that helps explain global and regional vegetation and biome changes [1], fractional vegetation cover (FVC) is commonly used in vegetation change analysis [2], climate prediction [3], soil erosion monitoring [4], drought monitoring [5], desertification assessment [6], hydrological modeling [7], agricultural monitoring [8], and other areas. Estimating the FVC within a region not only provides rapid insights into the effectiveness of regional ecological environment changes but also plays a crucial role in ecological environment monitoring [9]. Evaluation and monitoring of FVC will assist in assessing vegetation changes in rare-earth mining regions, safeguarding local ecosystems and fostering the sustainable development of environmental resources. These efforts will thereby profoundly influence the social and economic development of rare-earth mining regions.

Traditional vegetation coverage monitoring often employs conventional ground-based measurement techniques [10]. However, due to various limitations such as high cost, labor intensiveness, and time [11], this method is predominantly applicable to vegetation coverage detection in smaller areas. In contrast to traditional field measurements, remote sensing satellite technology offers a significantly more efficient detection approach [12].

Remote sensing technology offers broad coverage, extensive data collection, and rapid information acquisition, thereby facilitating economical and effective monitoring of larger geographical areas [13,14,15,16]. Consequently, with the ongoing advancements in remote sensing technology, FVC estimation using remote sensing data has become widely adopted in studies of spatio-temporal variations and driving factors [17]. Satellites like the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Medium Resolution Imaging Spectroradiometer (MERIS) have emerged as key data sources for examining long-term vegetation trends due to their wide coverage and high temporal resolution [18]. However, their coarse spatial resolution is often inadequate for capturing detailed spatial distribution, vegetation change patterns, and other critical information [19]. In comparison, the Landsat series offers a longer time series and finer spatial resolution, making it more suitable for FVC estimation in specific regions of interest. For instance, based on Landsat satellite observation data from 2004 to 2020, Chang et al. (2023) [20] studied the spatio-temporal variation characteristics of FVC in Harbin, China, and the results indicated that the average FVC declined as urbanization intensity increased, further quantifying the impact of urbanization on vegetation. Using Landsat images of arid Central Asia from 1989 to 2019, Dou et al. (2022) [21] identified the water extent of Lake Taitema and estimated the FVC of the surrounding areas, analyzing the changes in FVC and lake area before and after the implementation of the ecological water transmission project, along with its driving factors. The findings revealed that both precipitation and river inflow influenced local vegetation. Xiang et al. (2024) [22] analyzed changes in land use, FVC, and urban heat island intensity in Guangzhou from 1992 to 2022 using Landsat series data, and the results revealed a continuous decline in FVC, providing insights into the current status of urban land use and coverage, as well as the distribution and future trends of the heat island effect. Therefore, FVC estimation using Landsat series satellites can effectively capture vegetation growth characteristics, thus improving the precision of FVC estimation with fine resolution.

Rare-earth resources are crucial for the development of modern technology and are commonly used in the manufacturing of various high-tech applications, serving a vital strategic role in global economic development and technological advancement [23]. China holds extensive rare-earth element reserves, exhibiting a distribution pattern where heavier deposits are concentrated in the southern regions, while lighter ones are more prevalent in the north [24]. Ganzhou in Jiangxi Province hosts the largest ion-adsorption rare-earth deposits in China [25]. Dingnan County, situated within Ganzhou City, is a key region for rare-earth resource extraction in southern Jiangxi. The county possesses abundant mineral resources, which display intricate characteristics [26]. While early rare-earth mining brought rapid economic growth to the area, subsequent continuous and intensive mining activities exerted unprecedented pressure on the local environment, leading to numerous environmental and ecological challenges [27,28]. These crises include increased soil heavy metal content [29,30], reduced vegetation cover [31,32,33], dust pollution [34], and soil erosion [35]. These issues present significant threats to human health and the sustainable socio-economic development of the area.

Nevertheless, most existing studies on vegetation change in mining areas focus on large-scale opencast coal mines, while there has been limited focus on FVC variations in smaller county-level rare-earth mining areas, particularly in areas characterized by persistent cloud cover and rainfall in the south. For example, Zhang et al. (2023) [36] evaluated the vegetation restoration effect of the Shengli No. 1 open-pit coal mine dump using FVC differences and found that the restoration project initiated in 2016 yielded positive outcomes, highlighting the importance of timely and appropriate manual intervention for ecological restoration. Wang et al. (2021) [37] compared the growth trends of natural and restored vegetation to assess the condition of restored vegetation in Inner Mongolia coal mines, offering insights for ecological protection and mine development supervision. Hu et al. (2022) [38] utilized NDVI to monitor the vegetation reclamation effect of the Antaibao open-pit coal mine, providing a scientific basis for reclamation planning and land management. Xie et al. (2024) [39] examined vegetation disturbance and restoration patterns in abandoned open-pit mines around Dongting Lake, and the findings revealed that the restoration of various types of mining areas was contingent upon the extent of human intervention. Therefore, remote sensing technology is employed to explore FVC variations in rare-earth mining areas that have received limited attention in previous studies, to understand the spatio-temporal dynamics and development patterns of vegetation cover since the emergence of the rare-earth industry, and to analyze the reclamation efforts of typical southern Chinese rare-earth mining areas, exemplified by the Dingnan mining area, following the implementation of reclamation plans. This approach provides valuable insights for future restoration projects in mining areas worldwide and is essential for evaluating the ecological status of the Dingnan County mining area.

However, the intricate nature of the mining areas and their surrounding environments means that fluctuations in vegetation cover can be influenced by both increased human activities and shifts in natural conditions. Researchers, both domestically and internationally, have investigated the spatiotemporal evolution of vegetation and its driving mechanisms in various areas. Tian et al. (2024) [40] identified that, over the past two decades, precipitation, vegetation type, and soil type have been the primary determinants influencing vegetation cover in the Gansu region of the Yellow River Basin. Ma et al. (2024) [41] concluded that temperature and precipitation are the primary factors influencing changes in grassland vegetation cover in Xinjiang, while human activities may lead to changes in the types of grassland in the region. Chen et al. (2019) [42] discovered that the greening of forests and farmlands in China results primarily from large-scale afforestation and intensive farming, with human land use management being the main influencing factor for vegetation greening in China. It is evident that while mining activities are the largest contributing factor to changes in vegetation cover in mining areas, the impacts of natural and socio-economic factors, such as wind speed, precipitation, and surface temperature, on vegetation cover changes cannot be ignored.

Therefore, the objectives of this study are (1) to visualize the vegetation cover in Dingnan County over the past forty years and analyze the spatiotemporal characteristics of vegetation cover changes in this area. (2) To investigate the changes in FVC in Dingnan County, which will aid in tracking ecological transformations in regions impacted by rare-earth mining. (3) To identify the dominant factors influencing vegetation cover changes in mining areas through analysis of natural and socio-economic data in Dingnan County. This study offers a comprehensive analysis and valuable insights for vegetation reclamation in the Dingnan rare-earth mining area, and the research findings will provide a scientific foundation for the local government to formulate land reclamation plans and policies. At the same time, the adopted research method has global applicability and can be utilized to monitor the dynamics of vegetation restoration following mining area reclamation over an extended period and to monitor its recovery trajectory. This study contributes to the formulation of sustainable policies on a broader scale and provides scientific support for environmental protection and sustainable development decision-making in mining areas.

2. Materials and Methods

2.1. Study Area

Dingnan County (Jiangxi, China) situated at coordinates 24.62–24.94° N and 114.86–115.18° E, is positioned at the southernmost edge of Jiangxi Province (Figure 1) and covers an administrative area of 1321 km². The county experiences a subtropical seasonal climate, featuring hot, humid summers and cool, dry winters. The vegetation coverage rate in Dingnan County exceeds 82%, with natural vegetation unaffected by mining activities being well developed. The predominant vegetation types consist of warm coniferous forest, evergreen broad-leaved forest, bamboo forest, evergreen shrub forest, mountaintop coppice forest, and barren mountain shrub and grass [43,44]. The forest resources in the region are abundant, and the local vegetation remains stable across seasons, exhibiting minimal variation in NDVI.

This area is abundant in rare-earth reserves and boasts a long-standing history of rare-earth mining. Since the 1970s, the rare-earth extraction industry in this area has experienced swift growth. However, prolonged mining activities, particularly early techniques such as pond leaching and heap leaching, have resulted in the accumulation of large amounts of waste rock and tailings [45], severely polluting the local environment and impeding the recovery of natural vegetation while also posing health hazards to nearby residents [46,47]. Since the implementation of relevant mine management policies in 2011, many mining areas in Dingnan County have been closed, and mine reclamation efforts have progressively begun. However, the severe environmental problems caused by the ionic rare-earth mining industry continue to hinder ongoing environmental remediation and recovery efforts [48]. The mining of ion-type rare earth in Dingnan County has made great contribution to the development of the local economy and national high-tech industry; however, the substantial damage to vegetation resulting from prolonged mining, particularly its impact on the local ecological environment, cannot be overlooked.

2.2. Data Source and Preprocessing

2.2.1. Satellite Data Collection and Processing

This study utilizes satellite imagery data of Dingnan County in Ganzhou City from 1988 to 2023. Owing to the region’s placement in southern China, where cloudiness and rain are common, imagery was predominantly captured between December and February of the subsequent year to mitigate the effects of clouds and shadows. Additionally, the climate in southern China exhibits minimal seasonal variation, maintaining consistent vegetation cover throughout the year. This provides consistency in the spectral characteristics of ground features across different time periods, thereby ensuring a stable data foundation for the study. The satellite data were retrieved from the Landsat repository managed by the United States Geological Survey (USGS) (https://earthexplorer.usgs.gov, accessed on 11 November 2023): Landsat 5 (1988, 1995, 1997, 2002, and 2006), Landsat 8 (2013 and 2017), and Landsat 9 (2023). The spatial resolutions of the partial bands are 30 m, the panchromatic band is 15 m, and the thermal infrared band is 100 m. Radiometric calibration and FLAASH atmospheric correction are conducted using ENVI software 5.3 to mitigate potential interference in the satellite data [49]. Simultaneously, various sensors are calibrated to eliminate potential sensor differences [50,51]. Satellite data are processed through image registration and Mosaic to ensure the accuracy of data alignment and facilitate integration [51]. In addition, all satellite images utilized in this study had less than 10% cloud cover, ensuring the dataset’s reliability and high quality.

2.2.2. Driver Data Sets and Processing

According to the characteristics of the study area and previous studies, this study selected 8 related natural and man-made variables [52,53,54]. The meteorological data employed in this analysis, available from the National Meteorological Information Center (http://data.cma.cn, accessed on 13 November 2023), includes annual averages for surface temperature (X1), precipitation (X2), air temperature (X3), and wind speed (X4) spanning the period from 1988 to 2022. The nighttime light data (X5) were sourced from the annual China-wide dataset compiled by Wu et al. (2021) [55] (1995–2022). Information regarding Dingnan County’s Gross Domestic Product (GDP) (X6) and population (X7) for the years 2002 to 2023 was sourced from the statistical yearbooks published by the Dingnan County Bureau of Statistics (http://www.dingnan.gov.cn, accessed on 15 November 2023). Soil erosion dataset (X8) was derived from the 30 m annual soil water Erosion Dataset in China from 1990 to 2022 (http://www.scidb.cn, accessed on 12 November 2023).

The Natural Breaks (Jenks) method, applied within ArcGIS, was used to convert the seven continuous variables into categorical variables (

Table A1. Values of each factor type.

Year	X1	X2	X3	X4	X5	X6	X7	X8
2023		4	4	4	5	3	3
2017		1	4	1	4	3	1	1
2013	5	3	3	4	4	2	2	2
2006	2	5	3	5	3	1	1	3
2002	5	1	3	3	3	1	1	2
1997	3	5	2	2	1			5
1995	4	4	5	3	2			4
1988	1	2	1	4

). This approach efficiently lowers group variance while limiting the influence of human factors [56].

2.3. Methods

The research methodology employed in this study primarily focuses on estimating the vegetation cover in Dingnan County using the Dimidiate Pixel Model (DPM) applied to Landsat remote sensing imagery. Compared with empirical or alternative physical models, the DPM is an efficient and commonly used FVC estimation algorithm grounded in a robust theoretical framework and high accuracy [57]. However, the DPM also has some limitations, such as overlooking actual vegetation conditions and the multiple scattering interactions between soil and vegetation. Given the scarcity of corresponding ground-measured data in this study, along with the clear distinction between vegetation and non-vegetation in the study area, the DPM is selected as the method due to its relative independence from objective factors such as region and terrain, as well as its low dependence on ground-measured data, thereby minimizing errors arising from atmospheric conditions and soil types through appropriate calculations [58]. The driving factors are subsequently classified using ArcGIS software 10.2. Both environmental and anthropogenic influences on the relatively low vegetation coverage in Dingnan County for the years 1988, 1995, 1997, 2002, 2006, 2013, 2017, and 2023 were assessed. This analysis seeks to determine the primary factors influencing FVC within the county. Due to the lack of surface-measured data on vegetation in Dingnan County, the estimated FVC was not verified. The specific research process is depicted in Figure 2.

2.3.1. Vegetation Cover Calculation

FVC, representing the vertical projection of plant components (such as leaves, stems, and branches) compared to the overall surface area within a defined statistical region, is a widely employed metric in studies concerning ecological environments, land and water management, and climate variability [59]. The formula for calculating FVC is given by:

$$\mathrm{FVC}={\displaystyle \frac{(\mathrm{NDVI}-{\mathrm{NDVI}}_{\mathit{soil}})}{({\mathrm{NDVI}}_{\mathit{veg}}-{\mathrm{NDVI}}_{\mathit{soil}})}}$$

(1)

where:

$$\mathrm{NDVI}={\displaystyle \frac{NIR-R}{NIR+R}}$$

(2)

In Formula (2), the terms NIR and R correspond to the surface reflectance in the near-infrared and red spectral bands, respectively. NDVI_soil denotes the NDVI value for regions entirely lacking vegetation, such as bare soil, whereas NDVI_veg signifies the NDVI value for pixels fully covered by vegetation, representing pure vegetation. The calculation formulas for these two values are [60]:

$${\mathrm{NDVI}}_{soil}={\displaystyle \frac{{\mathrm{FVC}}_{\max }\times {\mathrm{NDVI}}_{\min }-{\mathrm{FVC}}_{\min }\times {\mathrm{NDVI}}_{\max }}{{\mathrm{FVC}}_{\max }-{\mathrm{FVC}}_{\min }}}$$

(3)

$${\mathrm{NDVI}}_{veg}={\displaystyle \frac{\left(1-{\mathrm{FVC}}_{\min }\right)\times {\mathrm{NDVI}}_{\max }-\left(1-{\mathrm{FVC}}_{\max }\right)\times {\mathrm{NDVI}}_{\min }}{{\mathrm{FVC}}_{\max }-{\mathrm{FVC}}_{\min }}}$$

(4)

The core principle of the DPM used for estimating FVC relies on the assumption that the information contained within a pixel is solely derived from vegetation and soil elements [61]. The cumulative NDVI value at 5% is taken as the NDVI_soil value, while the cumulative NDVI value at 95% is taken as the NDVI_veg value [62]. This implies that the maximum FVC (FVC_max) equals 100% and the minimum FVC (FVC_min) equals 0%. Therefore, the formula for calculating FVC can be expressed as:

$$\mathrm{FVC}={\displaystyle \frac{\left(\mathrm{NDVI}-{\mathrm{NDVI}}_{\min }\right)}{\left({\mathrm{NDVI}}_{\max }-{\mathrm{NDVI}}_{\min }\right)}}$$

(5)

Through this step, the calculation of FVC for Dingnan County from 1986 to 2023 was completed. To analyze the spatial distribution aspects and temporal variation characteristics of vegetation coverage in Dingnan County, this study classifies vegetation coverage into five levels by referring to previous classification criteria [53,63] and considering the actual conditions of the study area. The specific grades are presented in

Table A2. Vegetation cover scale.

FVC Value	Grade
0.0–0.2	Very low vegetation cover
0.2–0.4	Low vegetation cover
0.4–0.6	Moderate vegetation coverage
0.6–0.8	High vegetation cover
0.8–1	Very high vegetation cover

, where 0–0.4 denotes low vegetation coverage, usually corresponding to towns and mining areas with scarce vegetation; 0.4–0.6 represents moderate vegetation coverage, indicating moderate vegetation density; and 0.6–1 denotes high vegetation coverage, typically found in areas with dense vegetation such as forests [64].

2.3.2. Geodetector Model

Spatial differentiation is a core feature of geographical phenomena [65]. At present, various spatial models can be employed to capture the spatial relationships and variations between variables [66]. Geodetector, advanced by Wang et al. (2016) [67], is a model and tool designed to detect and explain the causes of spatial heterogeneity. Geodetector was chosen because it effectively identifies the spatial heterogeneity of a specific phenomenon within a given area and uncovers its underlying driving factors. As a result, it finds broad application in disciplines, including environmental science, earth sciences, and ecological studies [68,69].

The factor detector assesses the explanatory power of influencing factor X on the dependent variable Y [70]. This is measured using the q-value, which ranges from 0 to 1; a larger q-value signifies a greater degree of spatial differentiation explained by the independent variable on the dependent variable. The q-value undergoes a non-central F-test to determine significance levels [67].

$$\mathrm{q}=1-{\displaystyle \frac{{\displaystyle {\sum }_{h=1}^{L}Nh\sigma h^2}}{N{\sigma }^2}}$$

(6)

In Equation (6), q indicates the explanatory power of a factor in accounting for changes in relatively low-grade land area. Variable h represents the number of categories or classifications of Y or factor X, When the factors are divided into three categories, the variable h is 3, and N_h signifies the quantity of units classified within category h, whereas N represents the overall count of units throughout the entire area. σh² and σ² denote the variance of the Y value per unit of land area in category h and the entire region, respectively [71].

The interaction detector analyzes the relationships between natural factors, evaluates the interpretative strength of each of the seven variables, and determines whether the interplay between these factors is synergistic, weakening, or independent [72]. This process involves calculating the q-values of the two factors X1 and X2 for Y (q(X1) and q(X2)) and subsequently calculating the q-value for the interaction between the natural factors (q(X1∩X2)). The interaction q-value is then compared to q(X1) and q(X2).

3. Results and Discussion

3.1. Overall Evaluation of FVC in Dingnan County

Figure 3 illustrates the yearly variation in average vegetation coverage in Dingnan County. Over the 35-year period, the average vegetation coverage is 0.60. The highest value was recorded in 2023, while the lowest was observed in 2002. From 1988 to 2023, vegetation coverage in Dingnan County experienced an annual growth rate of 0.0014.

During this period, the area of high-level FVC initially declined before rising again, reflecting the success of vegetation restoration efforts in Dingnan County, as the vegetation ultimately returned to its original ecological state by the end of the monitoring timeframe. In the early years, vegetation coverage across various levels remained relatively stable, suggesting a steady state of vegetation, and mining activities had not significantly impacted vegetation coverage. The turning point appeared in 2002, after which the previously low level of vegetation cover exhibited a declining trend, suggesting an improvement in local vegetation cover. Since 2002, the government has progressively acknowledged the issue of unregulated mining in rare-earth mining areas and subsequently implemented relevant regulatory policies. At the same time, the introduction of new mining technology has significantly mitigated the damage to vegetation within the mining area. These two factors collectively facilitated the gradual restoration of vegetation in Dingnan County. Since 2006, the region characterized by minimal vegetation coverage (0–0.4) has experienced a persistent decrease, shrinking from 310.86 km² to 213.06 km². In contrast, the area of medium vegetation coverage (0.4–0.6) has steadily decreased since 2006, dropping to 10.60%. Simultaneously, the region exhibiting considerable vegetation coverage (0.6–1) has seen significant expansion since 2006, especially between 2017 and 2023, marked by an exceptional growth rate (

Table A3. Vegetation coverage of different grades in Dingnan County, 1988–2002.

Class	1988		1995		1997		2002
	Area (km2)	Percent	Area (km2)	Percent	Area (km2)	Percent	Area (km2)	Percent
Very Low (0–0.2)	161.81	12.25%	166.76	12.62%	169.86	12.86%	172.57	13.07%
Low (0.2–0.4)	137.58	10.42%	135.19	10.23%	134.25	10.17%	154.85	11.73%
Relatively Low (0–0.4)	299.39	22.67%	301.95	22.85%	304.11	23.03%	327.42	24.80%
Medium (0.4–0.6)	223.22	16.90%	206.25	15.61%	209.72	15.88%	236.96	17.94%
High (0.6–0.8)	343.25	25.99%	334.13	25.29%	345.51	26.16%	357.96	27.11%
Very High (0.8–1)	454.73	34.44%	478.91	36.25%	461.25	34.93%	398.26	30.15%
Relatively High (0.6–1)	797.98	60.43%	813.04	61.54%	806.76	61.09%	756.22	57.26%

).

Spatially, over the past four decades, regions with extremely high FVC have been concentrated in the high-altitude natural forests surrounding the mining sites. These areas, largely undisturbed by human activity, have preserved their natural condition (Figure 4, Figure 5 and Figure 6). Regions characterized by substantial vegetation coverage are predominantly located in low mountain and hilly terrains. Medium vegetation coverage areas were distributed in regions where hills transitioned to areas of human agricultural expansion. Areas with low FVC have been concentrated near urban infrastructure and mining zones, with particularly low levels observed in the central towns and primary mining extraction sites of Dingnan County.

3.2. Spatial and Temporal Distribution and Evolution of FVC

3.2.1. Changes in Vegetation Cover: 1988–2002

FVC maps and statistical data from 1988 to 2002 (Figure 4) clearly indicate that Dingnan County experienced a general decline in vegetation coverage over this time span. This decrease can largely be ascribed to the ongoing mining operations in the mining zones of Dingnan County from 1988 to 2002, leading to significant accumulation of substantial solid waste. Additionally, this trend reflects the ongoing enlargement of extraction activities within the region.

This observation is consistent with previous studies [73], indicating that the scale of rare-earth extraction in the southern region of Jiangxi expanded progressively during the late 1980s. However, mining technologies during the early stages of this expansion were still in developmental stages, and the in situ leaching method used initially often caused severe surface damage, resulting in the deterioration of aboveground vegetation [74]. This degradation made it difficult for the vegetation to recover, thereby creating significant challenges for subsequent restoration efforts. Through the examination of FVC reclassification maps in conjunction with the area and statistical information pertaining to different levels of FVC in Dingnan County between 1988 and 2002, the following observations are evident.

As illustrated in the statistical data (Figure 4e), the area corresponding to relatively low FVC levels (0–0.4) exhibited a steady increase between 1988 and 2002. The area of low FVC expanded from 299.39 km² (22.67%) in 1988 to 327.42 km² (24.80%) in 2002. In contrast, the area with relatively high FVC (0.6–1) experienced a decline, decreasing from 797.98 km² (60.43%) in 1988 to 756.22 km² (57.26%) in 2002. This trend indicates that mining operations in the region were consistently carried out, resulting in a continual expansion of these activities from 1988 to 2002 (Table A3).

The area corresponding to the relatively low vegetation coverage level (0–0.4) exhibited a consistent upward trend between 1988 and 1997, with no significant increase. Nonetheless, from 1997 to 2002, the increase in the low FVC area was markedly higher compared to the growth rate observed between 1988 and 1997. This shift is linked to the increasing demand for rare-earth metals after 2000 [75], which further accelerated the rapid expansion of mining activities.

3.2.2. Changes in Vegetation Cover: 2002–2013

Between 2002 and 2013, the area of relatively low-grade vegetation coverage (0–0.4) in Dingnan County’s mining areas experienced a consistent decline. In 2002, the extent of low FVC measured 327.42 km² (24.80%), but by 2006, this figure had diminished to 310.85 km² (23.54%) (

Table A4. Vegetation coverage of different grades in Dingnan County, 2002–2013.

Class	2002		2006		2013
	Area (km2)	Percent	Area (km2)	Percent	Area (km2)	Percent
Very Low (0–0.2)	172.57	13.07%	157.71	11.94%	152.28	11.53%
Low (0.2–0.4)	154.85	11.73%	153.14	11.60%	156.73	11.87%
Relatively Low (0–0.4)	327.42	24.80%	310.85	23.54%	309.01	23.40%
Medium (0.4–0.6)	236.96	17.94%	237.31	17.97%	268.96	20.37%
High (0.6–0.8)	357.96	27.11%	361.91	27.41%	375.62	28.44%
Very High (0.8–1)	398.26	30.15%	410.51	31.09%	367.02	27.79%
Relatively High (0.6–1)	756.22	57.26%	772.42	58.49%	742.64	56.23%

). An examination of the relevant information identifies three main factors contributing to the decline in low vegetation coverage during this timeframe: Policy Restriction: In 2004, the government initiated a mining enterprise in Ganzhou aimed at regulating the extent of rare-earth mining activities [76]. The policy was aimed at regulating extraction activities, thereby mitigating environmental impacts, particularly vegetation damage. Advancements in Mining Techniques: Starting in 2002, the implementation of “in situ leaching” technology was initiated in rare-earth mining areas, achieving full-scale adoption by 2006 [77]. The method involves the introduction of a leaching solution into the ore body, promoting the dissolution of rare-earth ions. Surface vegetation destruction is minimized, thereby achieving high recovery rates [78]. The period between 2005 and 2006 saw a substantial increase in rare-earth prices, which consequently prompted a growth in mining operations. However, this led to frequent illegal mining and mixed mining practices. In response, authorities initiated industry consolidation, closing numerous illegal mines [76,79].

From 2006 to 2013, the area of low-grade vegetation coverage continued to decline slightly, exhibiting a stable trend (Figure 5a–c). This stabilization period can be attributed to the gradual advancement of mine reclamation work in Dingnan County, starting in 2011. During this time, the local government introduced policies promoting green mining and increased regulation of rare-earth extraction activities. These measures led to the cessation of mining in certain areas, thereby facilitating a gradual recovery of low-grade vegetation coverage.

3.2.3. Changes in Vegetation Cover: 2013–2023

Between 2013 and 2023, Dingnan County underwent significant vegetation reclamation. A decline was noted in the expanse of land classified as having low quality (0–0.4), which diminished from 309.01 km² (23.40%) in 2013 to 213.06 km² (16.13%) by 2023 (

Table A5. Vegetation coverage of different grades in Dingnan County, 2013–2023.

Class	2013		2017		2023
	Area (km2)	Percent	Area (km2)	Percent	Area (km2)	Percent
Very Low (0–0.2)	152.28	11.53%	125.27	9.49%	117.88	8.93%
Low (0.2–0.4)	156.73	11.87%	115.66	8.77%	95.18	7.20%
Relatively Low (0–0.4)	309.01	23.40%	240.93	18.25%	213.06	16.13%
Medium (0.4–0.6)	268.96	20.37%	213.87	16.19%	139.91	10.60%
High (0.6–0.8)	375.62	28.44%	410.51	31.08%	325.17	24.62%
Very High (0.8–1)	367.02	27.79%	455.29	34.48%	642.46	48.65%
Relatively High (0.6–1)	742.64	56.23%	865.80	65.56%	967.63	73.27%

). Simultaneously, a steady increase in the region of relatively high-grade land was observed, rising from 742.64 km² (56.23%) to 967.63 km² (73.27%) (Figure 6). The FVC within and surrounding the mining area exhibited a consistent upward trend over time. This enhancement can be attributed to the successful execution of government policies and ongoing positive engagement from the community. Since the introduction of relevant policies in 2011, most mining activities in Dingnan County’s rare-earth mining areas had ceased completely by 2013, and reclamation efforts were fully initiated [80]. The degraded vegetation around the mining areas has progressively been restored to baseline levels.

3.3. Analysis of Driving Factors of FVC

To investigate the factors influencing vegetation coverage, a statistical analysis of the area of relatively low-grade vegetation (0–0.4) for each time period was conducted, serving as the response variable. The Geodetector model was employed to investigate the elements affecting variations in FVC throughout Dingnan County.

As shown in

Table 1. The effects of natural factors and human factors on vegetation coverage in Dingnan County; a higher q-value indicates that factors have a strong explanatory power for vegetation coverage.

Index	Dingnan County
	q-Value	p-Value	Sorting of q-Values
Surface temperature (X1)	0.666	1.000	5
Average rainfall (X2)	0.294	0.996	8
Temperature (X3)	0.945	0.286	3
Average wind speed (X4)	0.459	0.995	7
Night light data (X5)	0.773	0.740	4
GDP (X6)	0.947	0.189	2
Population (X7)	0.580	0.830	6
Amount of soil erosion (X8)	0.962	0.518	1

, the q-values for each factor indicate that the soil erosion amount in Dingnan County exerted the greatest influence, followed by GDP. As a measure of regional economic progress, GDP is strongly associated with mining operations in the Dingnan rare-earth mining region. It acts as a significant element affecting changes in the area of relatively low-grade vegetation. The amount of soil erosion partially reflects the mining activities in the region and has a significant impact on vegetation. Soil erosion limits the growth and resilience of local vegetation by directly causing the loss of topsoil. The deterioration of soil quality in this area is closely linked to mining activities, suggesting that local FVC is significantly influenced by mining activities. Meanwhile, the influence of annual average temperature ranked second to GDP, indicating that both natural and anthropogenic factors have influenced FVC in Dingnan County to varying degrees.

Our findings indicate that the interaction q-statistics for paired factors exceed those of the individual factors, implying that the combination of most indicators significantly accelerates or intensifies changes in the area of relatively low-grade vegetation in Dingnan County. This effect is especially pronounced for factors with q-values greater than 0.9 (Figure 7). Furthermore, the interplay among all factors is characterized by an amplification of the effects of two variables, with no indications of independent interactions that would reduce their influence. Current analysis indicates that annual average temperature, GDP, soil erosion amount and nighttime light values are key drivers influencing vegetation coverage in Dingnan County. Nighttime light values, which serve as an indicator of socioeconomic activity, also provide insights into the extent of land use in Dingnan. Extensive land use can alter the landscape complexity of the study area [81] and may exert considerable effects on the local ecosystem. Furthermore, temperatures exceeding a certain threshold can hinder vegetation growth in the study area [82], thereby affecting vegetation coverage.

From Table 1, it is found that the results obtained using geographic detectors are often associated with high p-values. Through discussion and reference to relevant data, we believe that if no statistical significance test is applied to the q-value, meaning the p-value is not considered, the q-value still retains clear physical significance [67]. Therefore, the conclusions drawn from the geographic detector exhibit a degree of reliability.

3.4. Analysis of FVC Variation in Typical Rare-Earth Mining Areas

3.4.1. Examination of Factors Influencing FVC Change in the Lingbei Mining Area

Although Dingnan County hosts numerous mining areas, more than 50% of the land remains undeveloped and minimally impacted by human activities. Therefore, in analyzing the driving factors in Dingnan County, these undisturbed areas may introduce certain deviations in the analysis of vegetation cover change in rare-earth mining zones. This variation in regional distribution may obscure the distinct ecological driving factors of mining zones and affect the accurate understanding and analysis of ecological changes in these zones. To further explore the key driving factors influencing vegetation change around rare-earth mining zones, the focus was placed on the Lingbei rare-earth mine, the most prominent ion-adsorption mining site in Dingnan County, in conjunction with other land use conditions. By comparing the driving factors influencing vegetation cover change between typical rare-earth counties and mining zones, the differences between the two and their distinct effects on FVC changes are revealed. The Lingbei rare-earth mine is situated in the northern part of Dingnan County. Between 1988 and 2002, the mining area exhibited a consistent expansion trend, suggesting frequent mining activities in Dingnan during this period and a continuous increase in the mining scale within the Lingbei area. A driving force analysis was conducted on its relatively low-grade FVC to explore the key factors influencing FVC changes in the mining area and the interactions between these driving factors.

The explanatory strength of different indices regarding vegetation coverage changes in the Lingbei mining area reveals that (

Table 2. The effects of natural factors and human factors on vegetation coverage in Lingbei rare-earth mining area; a higher q-value indicates that factors have a strong explanatory power for vegetation coverage.

Index	Lingbei Mining Area
	q-Value	p-Value	Sorting of q-Values
Surface temperature (X1)	0.976	0.360	1
Average rainfall (X2)	0.653	0.843	6
Temperature (X3)	0.897	0.236	2
Average wind speed (X4)	0.428	0.998	8
Night light data (X5)	0.726	0.605	5
GDP (X6)	0.861	0.464	4
Population (X7)	0.642	0.829	7
Amount of soil erosion (X8)	0.895	0.637	3

), with the exceptions of population, wind speed and average rainfall, most factors are significant contributors to spatial variability in the region. The highest explanatory power was exhibited by surface temperature (97.6%), followed by temperature (89.7%) and soil erosion amount (89.5%).

The relationship between land surface temperature and vegetation cover has long been a critical focus in heat island effect research [83]. Through photosynthesis, transpiration, and evapotranspiration, green vegetation significantly affects surface temperature, thereby effectively lowering surface temperatures. However, in rare-earth mining areas, the degradation of the natural surface structure due to mining activities leads to the appearance of bare ground, rendering the area more prone to solar radiation absorption, resulting in an increase in surface temperature. This temperature rise directly affects soil moisture and evaporation rates, negatively impacting the growth and survival of vegetation in the vicinity of the mining area. In addition, mining activities also release substantial heat, exacerbating the warming effect and contributing to local climate warming.

From the analysis in Figure 8, it is evident that the interaction between surface temperature and nighttime illumination markedly amplifies the influence of FVC in mining areas. Nighttime illumination reflects not only the prosperity of the area but also the mining operations in the region. At the same time, the interaction between surface temperature and precipitation has a notable amplifying effect compared to the individual impact of precipitation alone, with its influence being significantly greater than that of surface temperature on its own. These results indicate that the combined effect of precipitation and surface temperature on the FVC in rare-earth mining areas is considerable.

The analysis of primary influencing factors in the rare-earth mining zones of Dingnan County and Lingbei County reveals that vegetation cover changes in Dingnan County are predominantly shaped by soil erosion, GDP, and temperature, while in the Lingbei mining area, surface temperature, air temperature, and soil erosion play more significant roles. A key difference lies in the influence of GDP: in Dingnan County, where diverse economic activities extend beyond rare-earth mining, GDP-related activities more directly impact vegetation cover. Conversely, in the mining-centric Lingbei area, GDP exerts a more indirect effect, with surface temperature showing a more pronounced direct impact on vegetation coverage. This variation suggests that when focusing on specific rare-earth mining zones, the main influencing factors shift, with dominant factors often directly affecting surface vegetation.

3.4.2. Effects of Rare-Earth Mining on FVC

The findings highlight the impact of rare-earth mining activities on changes in vegetation coverage throughout Dingnan County. Since the late 1980s, the expansion of mining activities has exacerbated the increase in low-grade vegetation coverage, with the most severe destruction occurring in 2002 (Figure 9a). Although policy interventions during that time aimed to curb overexploitation, illegal mining continued due to the rising prices of rare-earth minerals. After 2013, Dingnan County strictly enforced national environmental protection policies, halting most mining activities and initiating large-scale reclamation efforts, leading to notable outcomes.

Over the past 35 years, areas with relatively low-grade FVC in Dingnan County remained stable in earlier years but began to decline following the implementation of relevant policies. In contrast, the Lingbei rare-earth mining area experienced an initial increase in low-grade FVC, followed by a decrease. As shown in the FVC trends for both regions (Figure 9b), Dingnan County’s mining scale expanded consistently from 1988 to 2006, with similar trends in relatively low FVC between 2013 and 2023, indicating that policy interventions effectively curtailed local mining activities. These regulatory actions impacted not only the immediate mining vicinity but also had broader effects, ultimately enhancing the overall vegetation coverage in the region.

3.5. Limitations and Future Work

This method of analyzing mining pollution through the calculation of changes in vegetation coverage around mining areas provides valuable insights into the severity of pollution and degradation, the progress of pollution control efforts, and land reclamation initiatives. However, several limitations exist:

• Inconsistencies in satellite data: While the Landsat satellite imagery was selected with relatively small time gaps, statistical inaccuracies still persist. Images prior to 2017 were obtained from Landsat 5, those from 2017 from Landsat 8, and those from 2023 from Landsat 9. The differences in satellite data products may affect subsequent statistical results, introducing certain limitations to the study.
• Environmental factors affecting images and image processing: The accuracy of results may be affected by conditions such as atmospheric interference, particulate matter, and various environmental elements. [84,85]. Cloud cover can alter spectral values, resulting in errors when measuring continuous variables such as vegetation biomass. The NDVI is especially prone to distortion due to atmospheric conditions such as cloud cover [86]. Meanwhile, processing raw satellite images of lower quality can introduce significant statistical errors. Therefore, it is essential to employ various analytical methods and data processing strategies to reduce the impact of inaccuracies. Addressing striping artifacts in satellite imagery is critical to maintaining accuracy.
• Lack of field data: While shifts in vegetation coverage can reflect the degree of degradation and pollution related to mining activities, especially concerning solid waste emissions, examining these changes over time and space does not completely reveal the underlying causes. Field survey data must be combined with statistical results to allow for more comprehensive analysis and accurate inferences. In addition, due to the lack of field data, the selection of driving factors in this study remains limited. Meanwhile, although existing application examples have demonstrated the reliability of DPM in FVC estimation, it is still necessary to evaluate its accuracy using field-measured data. The lack of truth-value data creates uncertainty in this analysis, which undermines the overall evaluation and interpretation of the findings.
• Discretization in Geodetector analysis: When using the Geodetector method for driving factor analysis, the factors must be discretized. Various discretization techniques can impact the results to a certain degree. Additionally, due to incomplete data for Dingnan County prior to 2000, the number of discrete variables is limited, which may lead to statistically insignificant results, thus influencing the overall findings.

Despite these limitations, the method of calculating vegetation coverage and generating reclassified vegetation coverage maps for mining areas continues to be an effective approach for representing vegetation changes. This enables the derivation of conclusions regarding the expansion of mining operations and the progress made in land reclamation initiatives. Visualizing the extent of mining damage provides a basis for local governments to formulate environmental remediation policies. Moreover, integrating vegetation coverage changes with historical satellite imagery facilitates a comprehensive assessment of pollution control efforts, reflecting the initiatives of local governments and mining companies in pollution remediation and land reclamation. This offers confidence and encouragement for future efforts.

In addition, using the Geodetector method to analyze the driving factors of vegetation coverage in Dingnan County and the Lingbei mining area allows for effective quantification of the influence of natural and human factors on FVC changes. The study results demonstrate that the Geodetector method is efficient in analyzing complex driving forces behind vegetation changes, which has significant implications for predicting future vegetation changes under evolving environmental conditions.

In future studies, the incorporation of higher-resolution satellite data and field measurements will be prioritized to overcome the shortcomings of this research. By improving the reliability of estimation methods and remote sensing satellites, the aim is to provide a more accurate basis for ecological assessment and offer greater scientific support for the ecological management and restoration of mining areas.

4. Conclusions

This study, utilizing the pixel dichotomy model and the Geodetector model, quantified vegetation coverage in Dingnan County, Jiangxi Province, from 1988 to 2023, examining the spatiotemporal changes in vegetation coverage over nearly four decades. Additionally, the study explored the primary driving factors influencing these changes. The findings indicate the following: (1). Dingnan County has exhibited a continuous upward trend in vegetation coverage, which reflects local governmental initiatives in mine management and land reclamation. (2). Throughout the last four decades, areas with high vegetation coverage have been predominantly located in forested regions distant from human activities, whereas areas around mining sites have typically shown lower vegetation coverage. (3). The interplay between environmental factors and FVC varies across different regions. In Dingnan County, changes in vegetation coverage have been shaped by both natural and human-induced factors, each exerting varying degrees of influence. GDP was identified as the dominant factor, while vegetation changes near the primary rare-earth mining areas were mainly influenced by surface temperature. According to the research results, it is recommended that mining managers enhance regulations on mining activities in these areas, adopt corresponding measures such as protective strategies and afforestation to facilitate the reclamation of local vegetation, and strengthen educational initiatives for local residents to encourage their participation in sustainable development efforts. This research facilitates the coordination of ion-type rare-earth mining and ecological protection, thus promoting sustainable development in the mining area while providing valuable scientific guidance for local governments to formulate more effective management and protection strategies for the mining ecosystem. Additionally, the methodology employed in this research possesses global applicability, aiding other mine managers in developing sustainable policies on a larger scale and providing a robust scientific foundation for informed decision-making regarding environmental protection and sustainable development in mining contexts.