Dynamic Remote Sensing Monitoring and Analysis of Influencing Factors for Land Degradation in Datong Coalfield

Abstract

Land degradation is one of the significant ecological and environmental issues threatening regional sustainable development. Datong Coalfield is located in an arid and semi-arid ecologically fragile area and is also an important energy base, the mining of coal resources and natural factors have caused serious land degradation problems. Therefore, dynamic monitoring and influencing factor analysis of land degradation in the Datong Coalfield is particularly important for land degradation prevention and land reclamation in mining areas. This study focuses on the Datong Coalfield, using remote sensing technology to dynamically extract soil erosion, net primary productivity of vegetation, land desertification, soil moisture content. Based on the Analytic Hierarchy Process (AHP), a comprehensive assessment model for land degradation was constructed to analyze the spatiotemporal evolution of land degradation in the Datong Coalfield from 2000 to 2021, and the influencing factors of land degradation were explored using a geographic detector. The results indicate that (1) from 2000 to 2021, the land degradation level in Datong Coalfield changed to mild degradation and non degradation, with the mild degradation area increasing by 30.48% and the non degradation area increasing by 13.9%, and spatially expanding contiguously from localized areas outwards. (2) Over the past 21 years, the land degradation situation in Datong Coalfield predominantly showed an improving trend, accounting for 69.11%, indicating an overall positive trajectory. However, 0.54% of the area experienced significantly intensified land degradation, scattered in the eastern and southwestern parts of the Datong Coalfield, which are areas requiring focused governance efforts. (3) Vegetation and land use are the main factors affecting land degradation in Datong Coalfield. At the same time, the influence of land use has gradually increased over the years, and the influence of vegetation and land use interaction is the highest in the two-factor interaction.

1. Introduction

Land degradation refers to the temporary or permanent decline in land productivity or its capacity to provide ecosystem services [1]. Under rapid socio-economic development, the irrational exploitation and utilization of land resources by humans coupled with population pressure have led to increasingly widespread and severe land degradation globally [2]. Land degradation has serious negative impacts on agriculture, environment, ecosystems, and poses a threat to socio-economic development. Consequently, it has garnered significant attention from environmental protection organizations, land management departments, and related institutions [3,4].

However, a consensus on the assessment and monitoring of land degradation has not yet been reached, particularly for studies at regional and larger spatial scales [5]. This lack of consensus is primarily manifested in two aspects: first, the construction of systematic and comprehensive monitoring indicator systems for specific regional types (such as ecologically fragile mining areas) is not yet mature; second, the analysis of long-term spatiotemporal dynamic evolution characteristics and driving mechanisms of land degradation at regional scales remains insufficient. The key to land degradation monitoring lies in integrating various parameter indicators to establish a land degradation assessment model, which requires determining monitoring indicators based on the manifestation characteristics of land degradation in the specific region. Existing research indicates that land degradation is often reflected in aspects like soil and vegetation. Scholars analyze dynamic changes by extracting and utilizing relevant indicators to characterize land degradation [6,7].

Remote sensing technology plays an important role in land degradation monitoring due to its advantages of wide coverage, high time efficiency and high cost-effectiveness. It can quickly and efficiently extract land degradation indicators in large areas and complete dynamic monitoring of land degradation [8,9]. Consequently, remote sensing-derived indicators such as the Normalized Difference Vegetation Index (NDVI), land surface temperature index, and desertification index have been used to determine land degradation status in numerous regions worldwide [10,11].

However, existing research on typical arid and semi-arid ecologically fragile mining areas like the Datong Coalfield often suffers from the following limitations [12,13]. (1) The selection of indicators often focuses on a single indicator (such as relying mainly on vegetation index), failing to systematically integrate key indicators that reflect the comprehensive degradation characteristics of mining areas. (2) The analysis of the dynamic evolution process of long-term series (such as more than 20 years) is insufficient, and is mostly limited to current status assessment or short-term comparative studies. (3) There is relatively little quantitative investigation into the driving factors influencing the spatial pattern and dynamic changes of degradation.

As a large comprehensive energy base in northern China, the Datong Coalfield has severe land ecological problems caused by large-scale and long-term coal mining [14]. However, research on the dynamic characteristics and influencing factors of land degradation in the Datong Coalfield is scarce. In view of the above research deficiencies, this study takes Datong Coalfield as a typical case area, (1) aims to build a regional land degradation remote sensing monitoring system: combining regional characteristics and existing research, systematically integrating four key indicators, namely Revised Universal Soil Loss Equation (RUSLE), Net Primary Productivity (NPP), Desertification Difference Index (DDI), and Surface Water Content Index (SWCI), and establishing a comprehensive land degradation index (LD) based on the analytic hierarchy process (AHP) to more comprehensively and objectively evaluate the land degradation status in the mining area. (2) Reveal spatiotemporal dynamic patterns: Utilizing remote sensing technology to dynamically extract the aforementioned indicators, we systematically analyze the spatiotemporal pattern evolution characteristics and change trends of land degradation in the Datong Coalfield over the 21-year period from 2000 to 2021. (3) Quantitatively analyze multi-factor driving mechanisms: Applying the geographical detector model, we quantitatively assess the independent impacts of natural and anthropogenic factors (vegetation, land use, topography, climate, population density, etc.) on land degradation. We place particular emphasis on detecting the interactions between these factors to uncover the underlying driving mechanisms.

This study is not only a detailed assessment of land degradation status in the Datong Coalfield but also, at the regional (mining area) scale, provides new methods and insights for understanding the land degradation processes in ecologically fragile mining areas through the construction of a systematic indicator system, long-term dynamic monitoring, and the analysis of driving factors. The research findings can provide a basis for land restoration, ecological protection, and the construction of green mines in coal mining areas.

2. Materials

2.1. Study Area

The Datong Coalfield is located in the northwest of Shanxi Province and is one of the six largest coalfields in Shanxi Province. Its geographical location is 112°30′~113°15′ E, 39°55′~40°12′ N, and it spans Yungang, Huairen, Zuoyun, Shanyin and other counties (Figure 1). The altitude ranges from 1026 to 1978 m, with an overall topographic feature of being higher in the southwest and lower in the northeast [15]. The region experiences a temperate continental monsoon climate characterized by short, warm, and rainy summers, as well as long, cold, and dry winters. The average annual temperature ranges between 4.6 °C and 6.8 °C, with annual precipitation measuring 380–460 mm, classifying it as an ecologically fragile arid to semi-arid zone. As a multi-period coalfield in the northern North China Coal Accumulation Area, the Datong Coalfield encompasses a coal-bearing area of approximately 1827 km². Its proven coal reserves reach 35 billion tons. Before 2000, the area was mainly open-pit mines and small-scale underground mines, with severe surface disturbances. After 2005, it was integrated into the Datong Coal Mine Group, forming a centralized and large-scale mining system. Since 2010, the policy of “mining and reclamation” has been implemented, and the efforts to replant vegetation and reclaim land have been intensified. Under such long-term, high-intensity mining, approximately 1687 km² of coal mining subsidence areas have formed [16,17]. In addition, cultivated land and grassland account for 82%, so some land degradation phenomena such as abandoned cultivated land and grass plantation degradation have occurred. It is precisely under the dual effects of ecological fragility and mining pressure that the Datong coalfield has become a typical land degradation area, with prominent problems such as vegetation degradation, land desertification, and soil erosion.

2.2. Data Source

This study selected Landsat 5 TM remote sensing images covering the Datong Coalfield for the years 2000, 2005, 2010, and Landsat 8 OLI images for 2015 and 2021. The image resolution of the data is 30 m, and the images with less cloud cover and clear imaging from July to September of the corresponding year were screened based on the GEE (Google Earth Engine) platform, after de-clouding, the images were synthesized by average and clipped according to Datong Coalfield. These were then used to calculate the normalized vegetation index, soil moisture content, and surface albedo.

Soil data comes from the Resource and Environmental Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn, accessed on 26 October 2024), with a spatial resolution of 1 km for soil texture type data, and from the National Tibetan Plateau Data Center (http://data.tpdc.ac.cn, accessed on 26 October 2024), with a spatial resolution of 1 km for Chinese soil organic matter content. They are used for soil erosion calculations.

Meteorological data are from the China Meteorological Data Network (http://data.cma.cn/, accessed on 29 October 2024). The monthly data set of China’s surface climate data and the monthly data set of China’s radiation in the corresponding years are selected for the calculation of soil erosion and vegetation NPP.

The DEM data comes from the Geospatial Data Cloud (http://www.gscloud.cn/, accessed on 11 October 2024). The ASTER GDEM data jointly developed by the Ministry of Economy, Trade and Industry (METI) of Japan and the National Aeronautics and Space Administration (NASA) of the United States has a spatial resolution of 30 m and is used for the extraction of soil erosion factors.

In addition, in the analysis of influencing factors, land use data comes from the Resources and Environmental Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn, accessed on 11 October 2024), and population density data comes from the World Pop Global Project Population Data.

3. Methods

3.1. Extraction of Land Degradation Monitoring Indicators

The key to establishing a land degradation monitoring indicator system lies in the determination of indicators. The indicators should reflect the characteristics of land degradation, be highly practical and facilitate data acquisition [18]. Based on the previous research on the establishment of a land degradation indicator system [5,19], combined with the main local environmental problems in Datong Coalfield, and considering the availability of remote sensing data to achieve dynamic updates, this paper selects the following four indicators to evaluate the land degradation problem in the study area.

3.1.1. Revised Universal Soil Loss Equation

Mining in mining areas will disturb the soil and underground soil layers, destroy the original soil structure, lead to increased soil erosion, and accelerate soil and water loss. Therefore, soil erosion is an important manifestation of land degradation in mining areas. The Revised Universal Soil Loss Equation (RUSLE) is used to calculate the soil erosion modulus [20,21]. The calculation process is shown in Formula (1).

$$RUSLE=R\times K\times LS\times C\times P$$

(1)

where $\mathrm{RUSLE}$ is the soil erosion amount (t·km⁻²·a⁻¹); $\mathrm{R}$ is the rainfall erosivity factor (MJ·mm·hm⁻²·h⁻¹·a⁻¹); $\mathrm{K}$ is the soil erodibility factor (t·hm²·h·hm⁻²·MJ⁻¹·mm⁻¹); $\mathrm{LS}$ is the slope length and slope factor; $\mathrm{C}$ is the vegetation cover factor; $\mathrm{P}$ is the soil and water conservation factor.

In actual calculations, the calculation of each factor can be modified accordingly with reference to existing research results and regional geographical conditions. The empirical formula proposed by Wischmeier et al. [22], widely used and suitable for the Loess Plateau region, is used to calculate the $\mathrm{R}$ value. The soil erodibility factor $\mathrm{K}$ value is calculated according to the EPIC model proposed by Williams et al. [23]. The LS factor is calculated using DEM data according to the method proposed by Liu et al. [24]. The method proposed by Shi et al. [25] for China is used to calculate the C factor. The P factor is often calculated using an assignment method, referring to existing research results [26] for assignment to complete the calculation.

3.1.2. Net Primary Productivity

Vegetation growth is closely related to land degradation in mining areas, and the severity of land degradation is directly affected by vegetation growth. Net Primary Productivity (NPP) is the amount of organic matter accumulated by plants per unit area per unit time, representing vegetation growth status and assessing land productivity. This study uses the improved CASA model [27,28] to calculate NPP. The model is based on the vegetation growth mechanism and is calculated through light energy utilization and photosynthetic active radiation. The specific calculation process is shown in Formulas (2)–(4).

$$NPP\left(x,t\right)=APAR\left(x,t\right)\times \epsilon \left(x,t\right)$$

(2)

where $\mathrm{APAR}\left(\mathrm{x},\mathrm{t}\right)$ is the photosynthetically active radiation absorbed by vegetation at pixel $\mathrm{x}$ during time $\mathrm{t}$ (unit: MJ/M²); $\mathsf{\epsilon }\left(\mathrm{x},\mathrm{t}\right)$ is the actual light use efficiency at pixel $\mathrm{x}$ during time $\mathrm{t}$ (unit: gC/MJ).

$$APAR\left(x,t\right)=SOL\left(x,t\right)\times FPAR\left(x,t\right)\times 0.5$$

(3)

where $\mathrm{SOL}\left(\mathrm{x},\mathrm{t}\right)$ represents the total solar radiation at pixel $\mathrm{x}$ during time $\mathrm{t}$ (MJ/M²); $\mathrm{FPAR}\left(\mathrm{x},\mathrm{t}\right)$ represents the fraction of photosynthetically active radiation absorbed by vegetation at pixel $\mathrm{x}$ during time $\mathrm{t}$; 0.5 represents the ratio of photosynthetically active radiation to total solar radiation.

$$\epsilon \left(x,t\right)=T{\epsilon }_1\left(x,t\right)\times T{\epsilon }_2\left(x,t\right)\times W_{\epsilon }\left(x,t\right)\times {\epsilon }_{max}$$

(4)

where ${\mathrm{T}\mathsf{\epsilon }}_1\left(\mathrm{x},\mathrm{t}\right)$,${\mathrm{T}\mathsf{\epsilon }}_2\left(\mathrm{x},\mathrm{t}\right)$, and ${\mathrm{W}}_{\mathsf{\epsilon }}\left(\mathrm{x},\mathrm{t}\right)$ represent the effects of high temperature, low temperature, and water stress on ${\mathsf{\epsilon }}_{\max }$ at pixel $\mathrm{x}$ during time $\mathrm{t}$; ${\mathsf{\epsilon }}_{\max }$ is the maximum light use efficiency.

3.1.3. Desertification Difference Index

Since the Datong Coalfield is located on the ecologically fragile Loess Plateau with intense mining activities, it is an important area for combating wind–sand desertification, and desertification is a significant manifestation of land degradation. The surface albedo-normalized difference vegetation index (Albedo-NDVI) feature space is constructed to extract the Desertification Difference Index (DDI) [29]. As shown in Formula (5).

$$DDI=-1/\left(a\times NDVI\right)-Albed$$

(5)

where $\mathrm{DDI}$ is the Desertification Index; $\mathrm{a}$ is the slope in the regression equation between $\mathrm{NDVI}$ and $\mathrm{Albedo}$.

$\mathrm{Albedo}$ is the surface albedo, calculated using the remote sensing model established by Liang et al. [30]. The process is shown in Formula (6).

$$Albedo=0.356B_{blue}+0.13B_{red}+0.373B_{nir}+0.085B_{swir1}+0.072B_{swir2}-0.0018$$

(6)

where ${\mathrm{B}}_{\mathrm{blue}}$ is the blue band, ${\mathrm{B}}_{\mathrm{red}}$ is the red band, ${\mathrm{B}}_{\mathrm{nir}}$ is the near-infrared band, ${\mathrm{B}}_{\mathrm{swir}1}$ is the short-wave infrared band, ${\mathrm{B}}_{\mathrm{swir}2}$ is the far-infrared band.

3.1.4. Surface Water Content Index

Soil water content reflects both the degree of land drought and the risk of land desertification, serving as an important indicator of land degradation, serving as an important indicator of land degradation. The Surface Water Content Index (SWCI) is selected to extract soil moisture [31]. The calculation process is shown in Formula (7).

$$SWCI=\left(B_{swir1}-B_{swir2}\right)/\left(B_{swir1}+B_{swir2}\right)$$

(7)

where ${\mathrm{B}}_{\mathrm{swir}1}$ is the short-wave infrared band, ${\mathrm{B}}_{\mathrm{swir}2}$ is the far-infrared band.

3.2. Construction of Land Degradation Index

3.2.1. Determination of Weights for Land Degradation Monitoring Indicators

To effectively assess land degradation, a comprehensive index needs to be constructed based on the land degradation monitoring indicators. The Analytic Hierarchy Process (AHP) is a mathematical method used for multi-criteria decision analysis [32], an effective method for quantitative analysis of non-quantitative events, suitable for integration with GIS technology for land degradation assessment [33]. The relative weights of the four extracted indicators were determined using a knowledge-based spatial decision support system, combined with prior knowledge and literature, to construct the judgment matrix for the four indicators (

Table 1. Judgment Matrix for Land Degradation Assessment.

Monitoring Indicators	DDI	RUSLE	NPP	SWCI
DDI	1	1/2	1/6	3
RUSLE	2	1	1/2	5
NPP	6	2	1	8
SWCI	1/3	1/5	1/8	1

).

According to the judgment matrix, the corresponding eigenvector $\mathrm{W}$ is obtained using the iteration method and Gaussian elimination. The calculation is shown in Formula 8.

$$W={\left(W_1,W_2,W_3,W_4\right)}^{\mathrm{T}}={\left(0.125,0.265,0.556,0.054\right)}^{\mathrm{T}}$$

(8)

where ${\mathrm{W}}_1,{\mathrm{W}}_2,{\mathrm{W}}_3,{\mathrm{W}}_4$ are the extracted land degradation monitoring indicators, corresponding to the Desertification Index (DDI), Soil Erosion Modulus (RUSLE), Net Primary Productivity (NPP), and Soil Water Content Index (SWCI).

AHP provides a mathematical method to determine the consistency of indicator comparisons, using the Consistency Index (CI) and Consistency Ratio (CR). The process is shown in Formulas (9) and (10).

$$CI={\displaystyle \frac{{\lambda }_{max}-n}{n-1}}$$

(9)

$$CR={\displaystyle \frac{CI}{RI}}$$

(10)

where $\mathrm{CR}$ is the Consistency Ratio, representing the consistency result of the constructed matrix; $\mathrm{CI}$ is the Consistency Index; ${\mathsf{\lambda }}_{\max }$ is the maximum eigenvalue of the judgment matrix; $\mathrm{n}$ is the number of indicators; $\mathrm{RI}$ is the Random Index.

After calculation, the $\mathrm{CR}$ of the above matrix is 0.021317 < 0.1, which is acceptable, thus obtaining the corresponding weights for each indicator.

3.2.2. Construction of Comprehensive Land Degradation Evaluation Index

Due to the inconsistency in the dimensions of the land degradation influencing factors, normalization is required. Analysis of the monitoring indicators shows they fall into two categories: one where the degree of land degradation increases as the factor value increases (Soil Erosion Modulus), requiring positive normalization; the other where the degree of land degradation decreases as the factor value increases (Net Primary Productivity, Desertification Index, Soil Water Content Index), requiring inverse normalization. The normalized calculation is shown in Formulas (11) and (12).

$$X_{scale}={\displaystyle \frac{X-X_{min}}{X_{max}-X_{min}}}$$

(11)

$$X_{scal{e}^{-}}={\displaystyle \frac{X_{max}-X}{X_{max}-X_{min}}}$$

(12)

where ${\mathrm{X}}_{\mathrm{scale}}$ and ${\mathrm{X}}_{{\mathrm{scale}}^{-}}$ are the results after positive and inverse normalization of the monitoring indicators respectively; $\mathrm{X}$ is the original data of the monitoring indicator; ${\mathrm{X}}_{\max }$ is the maximum value in the data; ${\mathrm{X}}_{\min }$ is the minimum value in the data.

After normalization, a comprehensive index model is used, performing weighted calculations according to the corresponding weights of each indicator to construct the comprehensive land degradation evaluation index, denoted as LD. The calculation is shown in Formula (13).

$$LD=0.125\times DDI+0.265\times RUSLE+0.556\times NPP+0.054\times SWCI$$

(13)

where LD represents the degree of land degradation, a higher LD value indicates a more severe degree of land degradation, and vice versa.

3.3. Trend Analysis

Trend analysis of raster data involves regression analysis of raster pixel variables that change over time to reveal the patterns of variable change [34]. Univariate linear regression can reduce the impact of extreme variables within a specific time frame on the analysis results [35]. This study uses univariate linear regression to analyze the trend of land degradation pixel by pixel. The calculation process is shown in Formula (14).

$${\theta }_{slope}={\displaystyle \frac{n\times {{\displaystyle \sum }}_{i=1}^n\left(i\times LD{I}_i\right)-{{\displaystyle \sum }}_{i=1}^{n}i{{\displaystyle \sum }}_{i-1}^{n}LD{I}_i}{n\times {{\displaystyle \sum }}_{i=1}^n{i}^2-{\left({{\displaystyle \sum }}_{i=1}^{n}i\right)}^2}}$$

(14)

where $\mathrm{n}$ is the number of years, which is 5 in this study; $\mathrm{i}$ is the annual variable, $\mathrm{i}=1,2,\dots \dots \mathrm{n}$; ${\mathrm{LDI}}_{\mathrm{i}}$ is the $\mathrm{LDI}$ value for the $\mathrm{i}$-th year. ${\mathsf{\theta }}_{\mathrm{slope}}$ represents the rate of change of $\mathrm{LDI}$ for a single pixel during the study period. When ${\mathsf{\theta }}_{\mathrm{slope}}>0$, it means that it has an increasing trend during the study period, otherwise it has a decreasing trend.

This study uses the $\mathrm{F}$-test to reveal the significance of the $\mathrm{LDI}$ change trend during the study period. The $\mathrm{F}$-test calculation formula is shown in Formula (15).

$$F=U\times {\displaystyle \frac{\left(n-2\right)}{Q}}$$

(15)

where $\mathrm{n}$ is the number of monitoring years; $\mathrm{U}$ is the sum of squares due to regression; $\mathrm{Q}$ is the sum of squares of residuals.

3.4. Geodetector

Geodetector, proposed by Wang Jinfeng et al., is a statistical method for quantitatively describing the spatial heterogeneity of geographical phenomena, analyzing relationships between variables, and revealing the driving forces behind them [36]. Its core idea is that if the independent variable has an impact on the dependent variable, then the spatial distribution of the independent variable and the dependent variable are similar, and the size of the influence can be measured by the spatial heterogeneity between variables [37,38].

This paper uses the factor detection module in the geographic detector to analyze the impact of various factors on land degradation. The impact is expressed by the spatial heterogeneity factor value. which is calculated as shown in Formula (15):

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

(16)

where $\mathrm{q}$ is the heterogeneity factor, with a value range of [$0,1]$. A larger $\mathrm{q}$ value indicates a greater influence of the factor on land degradation; $\mathrm{L}$ is the number of variable categories; $\mathrm{h}=1,2,\dots \dots \mathrm{L}$ is a specific category; $\mathrm{N}$ and ${\mathrm{N}}_{\mathrm{h}}$ are the number of units in the whole area and the number of units in the category respectively $\mathrm{h}$; $\mathsf{\sigma }$ and ${\mathsf{\sigma }}_{\mathrm{h}}$ are the variance of the whole area and $\mathrm{h}$ the variance of the category respectively.

Geodetector also features an interaction detector module, which can assess the interaction between different factors, i.e., determining whether the explanatory power for the dependent variable increases, decreases, or remains independent when two factors, ${\mathrm{X}}_1$ and ${\mathrm{X}}_2$, act together. First, the explanatory powers $\mathrm{q}\left({\mathrm{X}}_1\right)$ and $\mathrm{q}\left({\mathrm{X}}_2\right)$ of the two factors are calculated. Then, the explanatory power $\mathrm{q}\left({\mathrm{X}}_1\cap {\mathrm{X}}_2\right)$ after the interaction of the two factors is calculated. Finally, by comparing the values of $\mathrm{q}\left({\mathrm{X}}_1\right)$, $\mathrm{q}\left({\mathrm{X}}_2\right)$, and $\mathrm{q}\left({\mathrm{X}}_1\cap {\mathrm{X}}_2\right)$, the type of interaction between different factors can be determined [39]. The interaction types are shown in

Table 2. Types of interaction of factors.

Judgment Basis	Interaction Type
$q\left(X_1\cap X_2\right)<Min \left[q\left(X_1\right),q\left(X_2\right)\right]$	Nonlinear weakening
$Min\left[q\left(X_1\right),q\left(X_2\right)\right]<q\left(X_1\cap X_2\right)<Max\left[q\left(X_1\right),q\left(X_2\right)\right]$	Single-factor nonlinear weakening
$q\left(X_1\cap X_2\right)>Max\left[q\left(X_1\right),q\left(X_2\right)\right]$	Double factor enhancement
$q\left(X_1\cap X_2\right)=q\left(X_1\right)+q\left(X_2\right)$	Independence
$q\left(X_1\cap X_2\right)>q\left(X_1\right)+q\left(X_2\right)$	Nonlinear enhancement

.

4. Results

4.1. Spatiotemporal Characteristics of Land Degradation

The four land degradation indicators obtained through remote sensing inversion were input into the comprehensive land degradation evaluation index for calculation, yielding the land degradation results. Combined with field surveys and existing research, thresholds for different levels of land degradation were determined. The land degradation degree was divided into five levels: no degradation (0 ≤ LD < 0.35), mild degradation (0.35 ≤ LD < 0.45), moderate degradation (0.45 ≤ LD < 0.55), high degradation (0.55 ≤ LD < 0.65), and severe degradation (LD ≥ 0.65).

The spatiotemporal distribution characteristics of land degradation in Datong Coalfield from 2000 to 2021 are shown in Figure 2. The analysis shows that the land degradation in Datong Coalfield has changed significantly over the past 21 years. In 2000, land degradation was serious, and moderate and above degradation areas were significant. High and severe degradation were concentrated in the southwest and northeast (Shanyin County, Yungang District, etc.), moderate degradation was widely distributed, and non-degraded areas were scattered in the middle and north (Zuoyun County, Huairen City), while mildly degraded areas were mainly distributed in the periphery of non-degraded areas. In 2005, the highly and severely degraded areas decreased (obvious in Shanyin County, Zuoyun County, and Yungang District), and the mildly and non-degraded areas expanded to the central and eastern parts. It is worth noting that Shanxi Province had started coal integration work in 2004 and carried out pilot projects in Datong, so the improvement of land degradation in the periods of 2000–2005 and 2005–2010 was highly consistent with the implementation of Shanxi Province’s coal mine resource integration policy. The closure and integration of small coal mining enterprises under the policy effectively reduced surface disturbance, and the natural recovery of wasteland, grassland and vegetation in the region led to the transformation of degradation levels to mild and non-degraded. In 2010, 2015 and 2021, the land degradation situation further improved, and the scope of mildly degraded and non-degraded areas continued to expand. Highly and severely degraded areas still existed in the northeast and southwest (Yungang District, Youyu County, etc.), but the scope was reduced. Overall, the degree of land degradation remained at a low level, which was closely related to policies such as land reclamation and ecological restoration. However, from 2015 to 2021, land degradation intensified in some areas, which may be related to the intensified coal mining during this period and the continued development of surface collapse cracks in old goafs.

The proportion of each type of land degradation in Datong Coalfield was statistically analyzed (Figure 3). In terms of the proportion of each level of land, non-degraded land accounted for only 4.86% in 2000, and the proportion of land continued to rise thereafter, reaching 18.76% in 2021, a total increase of 13.90% in 21 years, indicating that the ecological quality of land in Datong Coalfield continued to improve. At the same time, the proportion of slightly degraded areas also increased year by year and was more significant, reaching a maximum of 46.05% in 2021, an increase of 30.48%, reflecting that the degree of land degradation is developing in a lighter direction. The proportion of moderately degraded areas showed a significant downward trend, from 55.29% in 2000 to 22.16% in 2021, a total decrease of 33.13%, which was the land degradation type with the largest change in 21 years. The proportion of highly degraded and severely degraded areas decreased in fluctuations, and the proportion of areas in the later period was relatively stable, accounting for 9.48% and 3.55% respectively in 2021.

This shows that the degree of land degradation in Datong coalfield has gradually changed to mild degradation and non-degradation over the years, demonstrating the role of coal mine resource management policies and ecological protection measures.

4.2. Analysis of Land Degradation Change Trends

The univariate linear regression analysis method was used to study the changing trend of land degradation in Datong Coalfield from 2000 to 2021, and the changing trend of land degradation was divided into 7 types as shown in

Table 3. Classification and area statistics of land degradation trends.

Classification Criteria	Changing Trend	Change Trend Category	Area/km2	Proportion/%
Slope ≤ 0.0005, A ≤ 0.01	Improving trend	Very significantly improved	13.667	0.70
Slope ≤ 0.0005, 0.01 < α ≤ 0.05		Significantly improved	235.565	12.12
Slope ≤ 0.0005, 0.05 < α		Not significantly improved	1094.355	56.29
−0.0005 < Slope < 0.0005	Not significant	Basically stable	354.501	18.23
−0.0005 ≤ Slope, 0.05 < α	Aggravating trend	Not significantly aggravated	235.456	12.11
−0.0005 ≤ Slope, 0.01 < α ≤ 0.05		Significantly aggravated	9.982	0.51
−0.0005 ≤ Slope, α ≤ 0.01		Very significantly aggravated	0.579	0.03

according to the change rate Slope and the significance test result α value. Combined with the spatial distribution of the changing trend of land degradation in Datong Coalfield (Figure 4), it can be found that the land degradation in most areas of Datong Coalfield showed an improvement trend (negative change rate), indicating that the overall improvement of land degradation in Datong Coalfield has been significant over the years, while the aggravation of land degradation (positive change rate) has a certain distribution in the northwest, north and northeast. The results of the significance test show that the area with an improving trend of land degradation is 1343.59 km², accounting for 69.11%, among which the largest area of the insignificant improvement area is 1094.63 km², accounting for 56.29%, which shows that although the land degradation in Datong Coalfield has improved overall, the improvement degree in most areas is not significant. The area with a trend of increasing land degradation is 246.02 km², accounting for 12.65%, of which the largest area with insignificant aggravation is 235.46 km², accounting for 12.11%. This shows that although land degradation has aggravated, the aggravation degree in most areas is limited. The area of basically stable area is 354.50 km², indicating that the land degradation status in some areas of Datong coalfield has been relatively stable for many years. At the same time, the analysis found that the area with extremely significant aggravation (0.03%) and significant aggravation (0.51%) accounted for a small proportion, mainly distributed in the east and southwest of the coalfield (such as Yungang District, Youyu County, and parts of Zuoyun County), and scattered in a dotted manner; while the extremely significant improvement (0.70%) and significant improvement (12.12%) are mainly distributed in the middle and south of the coalfield (such as Zuoyun County, Shanyin County, etc.), and distributed in continuous patches, which reflects the uneven changes in land degradation in Datong coalfield over the years, which may be related to the impact of local human activities.

In general, the land degradation in Datong Coalfield from 2000 to 2021 showed an improving trend. Although most of the improvements were not significant, the overall trend was positive. Only a few areas experienced aggravated land degradation and the degree was relatively mild. These areas should be paid special attention to in the future land degradation prevention and control. At the same time, the spatial pattern of land degradation trends in Datong Coalfield over the years was relatively complex and diverse, which may be closely related to factors such as human activities and natural environmental conditions in the region.

4.3. Analysis of Influencing Factors of Land Degradation

Based on the regional land degradation mechanism, this study comprehensively selected 8 factors in three categories: natural conditions, human activities, and ecological status. Among them, the natural background factors mainly selected topographic factors (elevation, slope, slope aspect) and climate factors (temperature, rainfall). Topographic factors affect vegetation growth and soil erosion by changing the distribution of water and heat; climate factors are important factors in the ecological vulnerability of arid and semi-arid areas, directly affecting vegetation growth and soil moisture. The human activity factors mainly selected land use types and population density to characterize the intensity of human activities, which can indirectly reflect the degree of disturbance of the surface caused by human activities such as mining and reclamation. The ecological status factor selected vegetation cover, which is the core representation indicator of land degradation and directly reflects the resilience of the land ecosystem. To ensure the consistency of the input data of the geodetector, this study classified land use, slope, and aspect according to actual research, and divided the remaining factors into 8 categories according to the natural break point classification method. The data were resampled into a 1 km × 1 km grid (a total of 1934 pixels) and input into the geodetector for single-factor detection and interaction detection.

The q value of the factor detection result indicates the influence of each factor on land degradation (Figure 5). The average q values of the five statistical periods are as follows: vegetation (0.71) > land use (0.42) > elevation (0.13) > population density (0.10) > rainfall (0.06) > temperature (0.03) > aspect (0.02) > slope (0.01). Vegetation and land use are the main driving factors of land degradation and have maintained a high influence for many years; elevation and population density have a certain influence on land degradation, but the influence is relatively low; while rainfall, temperature, aspect and slope have almost no influence on land degradation, and the average q value does not exceed 0.1.

From the changes in q values in different years, the factors that have the greatest impact on land degradation each year are vegetation, followed by land use. It is worth noting that the highest q value of vegetation appeared in 2000 at 0.76, and its q value decreased year by year over time; while the q value of land use increased year by year over time, reaching a maximum of 0.46 in 2021. This shows that land degradation in Datong coalfield is affected by natural factors (vegetation, elevation, rainfall, etc.) and human activity factors (land use, population density, etc.). Although the impact of natural factors is more significant, the influence of human activities is gradually increasing over time.

Interaction detection is used to evaluate the impact of driving factors on land degradation when they act together. The results of interaction over the years show (Figure 6) that the interaction shows nonlinear enhancement and double-factor enhancement, that is, the effect of any two factors interacting on the spatial differentiation of land degradation is greater than that of a single factor.

The influence of vegetation on land degradation after interaction with other factors is generally high, among which the influence of vegetation and land use interaction is the highest, reaching 0.769, and the influence of vegetation and slope interaction is also 0.733, both of which are greater than the influence of a single vegetation factor. This shows that vegetation status plays a key role in the land degradation process of Datong coalfield, and the interaction between vegetation and other factors significantly affects the trend of land degradation.

The interaction between land use and other factors also has a significant impact. In addition to the highest influence of land use interaction with vegetation, the influence of land use interaction with elevation is 0.436, and the influence of land use interaction with aspect is 0.387. In contrast, the degree of influence of factors such as aspect, slope, and rainfall after interaction with other factors is relatively small, indicating that these factors have relatively limited impact on land degradation when acting alone or when interacting.

In summary, the interaction between vegetation and land use and each factor has a strong impact on land degradation, that is, vegetation and land use are the key factors affecting land degradation in Datong coalfield, and the combined impact of natural factors and human activities on land degradation is significantly enhanced.

5. Discussion

The Datong Coalfield, located in northern Shanxi, is a key ecological conservation area both provincially and nationally. Characterized by a fragile natural environment and frequent human activities (such as mineral resource mining and grazing), it exhibits pronounced land degradation phenomena including soil erosion, desertification, and vegetation degradation. This study employs dynamic remote sensing technology to construct a multi-dimensional comprehensive evaluation system, implements long-term land degradation monitoring, and applies geographical detectors to quantitatively analyze driving factors, achieving innovative progress in land degradation research for the Datong Coalfield.

Previous land degradation assessments in mining or arid and semi-arid regions often over-relied on single indicators, failing to capture the compound characteristics of mining-area degradation. This study integrates four indicators—Revised Universal Soil Loss Equation (RUSLE), Net Primary Productivity (NPP), Desertification Difference Index (DDI), and Surface Water Content Index (SWCI)—to establish a comprehensive land degradation evaluation system.

Compared with single indicators or simple combination methods, this approach systematically and comprehensively characterizes the multi-dimensional state of land degradation in the Datong Coalfield. It reveals an overall trend of degradation transitioning from “moderate/high” to “mild/non-degraded” levels. The improvement based on the comprehensive indicators can better explain the overall effectiveness of various land reclamation and ecological restoration projects implemented in the study area, avoiding the misjudgment of a single indicator.

The existing studies on spatiotemporal evolution of land degradation mostly focuses on the analysis of the current pattern or the comparison of several periods of results, but it is difficult to accurately reflect the volatility of the degradation process. This study uses a time series analysis method based on the pixel scale to clearly depict the spatiotemporal trend of land degradation over the years. Although most areas show an improvement trend, there is still a 12.65% trend of aggravation, which provides a direct scientific basis for “zoning policy and precise governance”.

At present, there are few quantitative studies on the driving factors of land degradation. This study quantified the independent impact of each factor based on the geographic detector and revealed the intensity of the interaction between the factors. The study obtained new findings: Land degradation is co-driven by natural and anthropogenic factors, with the vegetation-land use interaction exhibiting the strongest explanatory power. The influence of land use on degradation has gradually increased over time. The quantitative characterization of driving factors provides scientific support for optimizing governance strategies.

6. Conclusions

This study addresses the land degradation issues in the Datong Coalfield by constructing a comprehensive land degradation index (LD) that integrates Revised Universal Soil Loss Equation (RUSLE), Net Primary Productivity (NPP), Desertification Difference Index (DDI), and Surface Water Content Index (SWCI) using remote sensing technology. It systematically analyzes the spatiotemporal evolution patterns and driving mechanisms of land degradation in the Datong Coalfield from 2000 to 2021. The main conclusions are as follows.

Over the 21-year period, land degradation in the Datong Coalfield exhibited a stepwise decline. The combined proportion of mild degraded and non-degraded areas increased from 20.43% to 64.81%. The spatial pattern presents a “core-periphery diffusion” pattern, with mildly and non-degraded areas expanding from the central part (Zuoyun County, Huairen City) to the periphery, while the degradation center of gravity has shifted from the southwest-northeast continuous distribution (Shanyin County, Yungang District) to sporadic contraction.

Land degradation showed improvement trends in 69.11% of the area (with 12.82% significantly improved), but 12.65% exhibited continued intensification (0.54% significantly intensified). Critically, the extremely significant intensification zones are scattered in the eastern and southwestern regions (Youyu County, Yungang District) and should be prioritized for targeted control. Future efforts should strengthen dynamic monitoring of non-significantly improved areas while consolidating and enhancing restoration outcomes through sustained ecological management measures.

Vegetation and land use emerged as the two core driving factors. Their interactive effect (q = 0.769) demonstrated the strongest explanatory power. All driving factors exhibited nonlinear enhancement or bifactorial enhancement, indicating that land degradation results from the combined effects of multiple factors. Temporally, vegetation’s influence decreased while land use impact steadily increased, reflecting the growing regulatory role of human activities.

Through the multi-indicator comprehensive assessment system, remote sensing dynamic monitoring, and quantitative analysis of driving factors, this study reveals the degradation process in the Datong Coalfield and the spatiotemporal mechanisms of “natural-anthropogenic” interactions, providing support for mine ecological restoration. However, this study also has certain limitations. Human activity factors (such as mining intensity, ecological restoration projects, etc.) were indirectly represented only through land use, failing to precisely distinguish the impacts of different human activities. Subsequent research should refine driving factor analysis with mining and ecological protection data. The study relied solely on remote sensing for degradation characterization without quantifying the feedback mechanism between vegetation restoration and soil properties during degradation. In the future, it is necessary to integrate multi-dimensional data such as soil organic matter to deepen the mechanism analysis.