Ecosystem Health Assessment of Coal Mining Subsidence Wetlands Using the DPSIR Model: A Case Study in Yingshang County, China

Abstract

Coal mining in the eastern Huaihe Plain has led to land degradation and hydrological disturbances, transforming terrestrial ecosystems into a complex of terrestrial and aquatic systems. These changes significantly impact regional ecological processes, structure, and functions. Hence, assessing the health condition and restoring the degraded subsidence wetlands efficiently have become urgent issues in coal resource-based cities. This research developed an ecosystem health assessment model for mining subsidence wetlands using the Driving Force–Pressure–State–Impact–Response (DPSIR) framework, with a focus on the subsidence wetlands of Yingshang County, Anhui Province. The assessment findings indicated that the wetland ecosystem was in a sub-healthy condition, with a health score of 0.51. Specific scores for the subsystems “Driving Force”, “Pressure”, and “State” were 0.584, 0.690, and 0.537, respectively, indicating that these subsystems were categorized as healthy, very healthy, and sub-healthy. In contrast, the scores for the “Impact” and “Response” subsystems were 0.076 and 0.093, both falling within the very poor (V) status. Weight analysis of the indicators revealed that the regional development index (Cp1), mining subsidence disturbance intensity (Cp2), aggregation index (Cs3), diversity index (Cs4), and wetland conservation rate (Cr1) significantly affected wetland ecosystem health. Taking into account both the health assessment results and the specific environmental conditions of the study area, this research recommends restoration strategies and the preservation of wetland ecosystems. The findings from this study can provide a basis for governmental bodies to create specific strategies and policies aimed at the conservation and management of subsidence wetlands.

1. Introduction

The mining industry has long faced environmental challenges due to coal mining, despite coal being the most widely distributed conventional energy source globally [1]. While coal provides raw materials and energy for human beings, it also produces negative ecological and environmental effects such as land excavation, suppression, and subsidence [2].

Subsidence wetlands are found in mining regions with high groundwater levels, including areas in eastern China [3,4], central Europe, the central United States, and eastern Australia [5,6]. Subsidence wetlands are a special type of artificial wetland where the original terrestrial ecosystem is gradually replaced by an aquatic ecosystem [7]. Subsidence wetlands are commonly regarded as a significant consequence of mining activities in resource-dependent cities, as they are associated with the destruction of cultivated land, decreased vegetation productivity, and increased soil erosion. However, it has been demonstrated that they could be regarded as potential green infrastructure (GI), providing ecosystem services and promoting the well-being of humanity [8].

In China, it is estimated that approximately 0.20 to 0.33 hectares of land experience subsidence annually for every 10,000 tons of coal mined underground [2]. In coal mining cities, underground coal mining is considered a significant factor in shaping wetland landscape patterns. However, under the continuous disturbance of underground coal extraction, mining subsidence wetlands are rapidly expanding. Simultaneously, urbanization and land reclamation have led to a substantial reduction in the area of natural wetlands. Numerous studies have shown that both human activities and natural factors influence wetland changes from the perspectives of hydrology, land use, and other processes [9,10]. Therefore, mining subsidence wetland ecosystems exhibit significant instability, with wetland habitat networks becoming increasingly fragmented.

To address these challenges, restoring damaged mining subsidence wetlands and improving wetland network structure have become priorities in protection and restoration efforts. Assessing wetland ecosystem health is essential for effective protection and restoration, aiding in the conservation and management of these critical habitats [11]. Wetland health assessment is essential for three reasons: (1) it reflects the wetland’s current condition, (2) it aligns with ecological processes, and (3) it illustrates how human activities interact with the wetland system. Various models and assessment methods have been used by researchers to evaluate wetland health. To date, wetland ecosystem health assessment has primarily focused on two aspects: First, researchers initially assess wetland health by studying indicator species, populations, or communities [12,13]. Second, considering wetlands as a comprehensive ecosystem of environment–society–economy, selecting key indicators and their weight coefficients that characterize the main features of the wetland ecosystem, and constructing an assessment index system to comprehensively reflect the health status. Duan et al. [14] conducted a survey on the socio-economic conditions, natural environment, and river water quality in the Chaohu Lake basin. A comprehensive approach for wetland ecosystem research has been suggested by Ritesh Kumar et al. [15], highlighting the importance of considering the interactions between ecological and social systems. Moreover, the PSR (Pressure–State–Response) model, along with its extended versions, DPSIR (Driving Force–Pressure–State–Impact–Response), are commonly used frameworks to depict the interactions between society and the environment. These frameworks employ a structured approach to understand the complexity of environmental issues and elucidate how various factors interact with socio-economic systems. Therefore, it has been applied in assessments of wetlands from different perspectives, such as risk and vulnerability [16], ecological security [17], and ecosystem health [18].

In the comprehensive assessment framework, numerous indicators are employed, and various models or methods have been applied worldwide for wetland research. While extensive studies have been conducted on the ecosystem health of inland wetlands [19], coastal wetlands [20], and wetland parks [21], subsidence wetlands—a special type of wetland formed by coal mining activities—have received relatively less attention. Hou et al. [22] assessed the impact of mining activities on mining subsidence wetlands through ecosystem service value measurements. Wang et al. [23] constructed a value assessment model based on cost–benefit analysis principles to evaluate the value of artificial wetlands created by mining subsidence. The research methods exhibit the following characteristics: First, some methods rely on data that are specific to the study area, which can pose challenges for other researchers in obtaining comparable data when applying these methods to different regions. Second, the determination of assessment indicator weights is crucial. By combining both subjective and objective weighting methods, information loss can be minimized, and the results can better reflect reality. Additionally, wetland health assessments typically yield numerical health grades, which often lack sufficient integration with spatial distribution patterns. Furthermore, there is a lack of in-depth analysis for the management and development of wetlands.

This study aims to develop a DPSIR assessment framework for evaluating the ecosystem health of subsidence wetlands, taking Yingshang County in Anhui Province as the example. The specific objectives include the following: (1) to apply the DPSIR model to create a comprehensive health index system for subsidence wetlands; (2) to compute the wetland health index and examine how driving forces contribute to pressures in coal resource-based cities, thus influencing the health of wetlands; and (3) to propose spatial planning strategies for the mining subsidence wetlands. Ultimately, this study aims to offer practical solutions for the ecological restoration of subsidence wetlands, fostering the sustainable, long-term development of both the regional ecology and economy.

2. Materials and Methods

2.1. Study Area

Yingshang County (115°56′ E–116°38′ E, 32°27′ N–32°54′ N), situated in north-western Anhui Province, falls under the administration of Fuyang City, Anhui Province (Figure 1). The county is characterized by a typical alluvial plain in the Huaibei region, with no hills or mountains and a flat terrain. The area features a crisscrossing network of ditches and rivers, forming a fluvial plain landscape. Yingshang County experiences a semi-humid monsoon climate, with an average annual precipitation of 926.1 mm over several years. The mining subsidence areas in Yingshang County are primarily concentrated in the north-east part of the county, including five townships, Jiangkou, Gucheng, Chenqiao, Digou, and Xieqiao, with a total area of 367.10 km². This region is a major coal-producing area in Huaibei and one of the 14 key coal bases developed under China’s national plan.

Yingshang County is abundant in coal resources and was recognized as a “resource-based city” by the State Council in 2013. However, coal mining has led to increasing ecological and social issues (Figure 2). Large areas of mining subsidence wetlands have formed, causing the destruction of farmland and infrastructure. These issues have severely affected the local living environment. In recent years, wetland resources have been given more attention by governments at all levels in terms of protection and sustainable utilization. The mining subsidence wetlands formed by the Xifei River, Ji River, Xieqiao Coal Mine, and Liuzhuang Coal Mine were approved as a pilot national wetland park by the State Forestry Administration in 2008. They were officially designated as the Anhui Digou National Wetland Park in 2016, with a total area of 3085.73 hectares. As of 2023, four mining areas have caused the formation of mining subsidence wetlands: Kouzi East Mine, Liuzhuang Coal Mine, Xieqiao Coal Mine, and Banji Coal Mine. The total area of these wetlands is 1513.26 hectares, accounting for 4.12% of the study area. Meanwhile, Xieqiao Coal Mine and Liuzhuang Coal Mine are still in normal production. Therefore, the scope of mining subsidence wetlands will continue to expand.

2.2. Data Source

The ecosystem health assessment focuses on the environmental issues of mining subsidence wetlands and constructs a health index system at the landscape scale to determine the condition of the wetland ecosystem. The data utilized in this study are as follows (

Table 1. Indicators for ecosystem health assessment in subsidence wetlands.

Type	Data	Resolution	Time	Source
Land use	Land use	30 m	2023	China Center for Geographic Information and Research (https://www.resdc.cn/, accessed on 19 September 2024)
Wetland category	LANDSAT-8 (OLI/TIRS)	30 m	2023	United States Geological Survey (USGS) (https://earthexplorer.usgs.gov/, accessed on 19 September 2024)
Environmental	Precipitation	1 km	2023	National Tibetan Plateau Data Center (https://data.tpdc.ac.cn/home, accessed on 17 March 2025)
	Digital Elevation Model (DEM)	30 m	2023	Geospatial Data Cloud (http://www.gscloud.cn/, accessed on 19 September 2024)
	Normalized Difference Vegetation Index (NDVI)	30 m	2023	China Center for Geographic Information and Research (https://www.resdc.cn/, accessed on 19 September 2024)
Socio-economics	Population	1 km	2023	LandScan (https://landscan.ornl.gov/, accessed on 17 March 2025)
	Road	1:3000	2023	Open Street Map (https://www.openstreetmap.org/, accessed on 19 September 2024)
	Wetland Conservation Area	1:3000	2019	Master Plan for Anhui Digou National Wetland Park (2019–2023)
Planning	Subsidence areas	1:3000	2023	Planning documents

):

(1) Land use data were obtained from the China Center for Geographic Information and Research (https://www.resdc.cn/, accessed on 19 September, 2024) at a spatial resolution of 30 m. These data were used to classify land cover into categories such as farmland, forest land, grassland, wetlands, construction land, and unused land. Landsat 8 (OLI/TIRS) images from 10 June 2023 were obtained from the USGS to extract wetland landscape information. Following the Chinese wetland classification standards, the wetlands in the study area were categorized into natural rivers, artificial canals, reservoirs, ponds, artificial lakes, and mining subsidence wetlands.

(2) Environmental data: Precipitation, Digital Elevation Model (DEM), and Normalized Difference Vegetation Index (NDVI) data were sourced from public datasets provided by the National Tibetan Plateau Data Center [24], Geospatial Data Cloud, and the China Center for Geographic Information and Research. The accuracy of these datasets has been validated.

(3) Socio-economic data: The population and road network data are from LandScan and Open Street Map. The Wetland Conservation Area is acquired from the Master Plan for Anhui Digou National Wetland Park (2019–2023) during the investigation with the local government.

(4) Planning data: mineral-related data were derived from the Geological Environmental Protection and Land Reclamation Plan for Mines.

2.3. Methodology

The DPSIR model was employed to evaluate the ecosystem health of subsidence wetlands, emphasizing four main components: (1) Constructing the DPSIR conceptual framework. (2) Identifying and selecting suitable assessment indicators. (3) Determining the comprehensive weights of these indicators using a combination of the Analytic Hierarchy Process (AHP) and the CRITIC method. The wetland ecosystem health level was then evaluated using a combined evaluation approach, providing a quantitative standard for assessing the health of subsidence wetlands. (4) The DPSIR model was also employed to analyze the ecosystem health index and its spatial distribution.

2.3.1. DPSIR Model

The DPSIR (Driving Forces–Pressure–State–Impact–Response) model builds on the PSR (Pressure–State–Response) model, which analyzes the pressure–state–response relationship in environmental issues. By incorporating “driving forces” and “impacts”, DPSIR enables a more thorough analysis of socio-environmental dynamics [24]. This framework categorizes and evaluates various factors to better understand the causal relationships between watershed ecosystem health and human socio-economic activities [25], which are bidirectional.

The DPSIR model has been widely applied in ecosystem health assessments due to its high versatility. It consists of five subsystems: (1) “Driving Forces”, which are the fundamental factors and primary motivations that drive the natural succession or degradation of wetland ecosystems, including both human activities and natural factors; (2) “Pressure”, which refers to the extent to which human economic and social activities affect wetland ecosystems; (3) “State”, which represents the condition of the ecological environment as influenced by pressure, particularly changes in landscape structure and function; (4) “Impact”, which refers to the effects of changes in the wetland ecosystem state on human survival and development; and (5) “Response”, indicating the management mechanisms’ reactions to changes in the ecosystem. These five factors are causally related and collectively express the coupled relationships affecting wetland ecosystem health (Figure 3). The DPSIR model is widely applicable in environmental assessment and governance. It helps policymakers and managers understand the root causes of environmental problems, thereby enabling effective response measures.

Population growth and environmental changes are long-term drivers that influence the spatial distribution and health of wetlands, contributing to urban expansion and the loss of ecological spaces [26]. In addition, the exploitation of coal resources and urban expansion have intensified the enormous pressure on wetland ecosystems. Under this pressure, wetland ecological structure and function will gradually be destroyed and ecological diversity will be lost. This results in the deterioration of wetland ecosystem services, including functions like rainfall absorption and flood regulation. Simultaneously, coal mining-induced subsidence wetlands will continue to expand, further exacerbating the loss of arable land and damaging infrastructure. To protect and restore these wetlands, governments and environmental agencies typically demarcate areas for ecological restoration and propose targeted strategies. These actions also influence the four previously mentioned components, creating a positive feedback loop that helps address ecological issues and fosters sustainability.

2.3.2. Determination of the Assessment Indicator System

Accurately evaluating the health of subsidence wetlands requires an effective and well-structured assessment indicator system. At present, however, a standardized and universally accepted set of indicators for assessing wetlands affected by mining subsidence has yet to be established. Assessment indicators should be effective and scientifically sound, reflecting the study area’s specific characteristics while offering a comprehensive and systematic evaluation of its health. Initially, relevant indices from the existing literature were compiled. Then, based on criteria such as data availability, quantifiability, objectivity, representativeness, independence, and regional relevance, overlapping or conflicting indicators were combined or removed, while those with frequent usage and substantial content were selected. Following these guidelines, 15 indicators were chosen based on the DPSIR model, as outlined in

Table 2. Assessment indicators for evaluating ecosystem health in subsidence wetlands.

Target Layer	Criterion Layer	Indicator Layer		Meaning
Ecosystem health (A)	Driving Forces (B1)	Cd1	Population density	The driving effect of population size on wetland ecosystem health, calculated as population per unit area (persons∙hm−2).
		Cd2	Elevation	The occurrence of severe surface subsidence and changes in surface landscape.
		Cd3	Annual average precipitation	Regional precipitation, reflected by annual precipitation.
	Pressures (B2)	Cp1	Regional development index	The degree of human land use and development pressure on wetlands, indicated by the proportion of urban and agricultural land within the total land area.
		Cp2	Mining subsidence disturbance intensity	The pressure of mining subsidence on wetland space, calculated as the ratio of subsidence area to township area.
		Cp3	Road density	The pressure of transportation on wetland environments, calculated by the road network length (km)/area of the region (km2).
		Cp4	Industrial disturbance	Industrial and mining land use on the sustainable development of wetland space, indicated by the distance between mining land and wetlands.
	State (B3)	Cs1	Normalized Difference Vegetation Index (NDVI)	The vegetation cover status of the study area.
		Cs2	Landscape Shape Index (LSI)	The complexity of patch shapes; higher values suggest greater landscape heterogeneity.
		Cs3	Aggregation Index (AI)	The connectivity between landscape patches; higher values indicate greater landscape connectivity.
		Cs4	Shannon Diversity Index (SHDI)	The diversity of wetland landscapes; higher values indicate greater landscape diversity.
	Impact (B4)	Ci1	Mining subsidence wetland rate	The proportion of the mining subsidence wetland area within the study region.
		Ci2	Hydrological regulation Index	The flood storage and water supply functions of wetland ecosystems, calculated as the sum of pond and ditch areas divided by the total study area.
	Responses (B5)	Cr1	Wetland conservation rate	The level of wetland protection, determined by the ratio of conserved areas to the total study area.
		Cr2	Patch Density (PD)	The degree of landscape fragmentation; higher values suggest lower wetland health.

.

1.

Driving Forces subsystem

“Driving forces” are the fundamental factors and primary motivations that drive the natural succession or degradation of wetland ecosystems. After reviewing previous studies, three indicators were selected to reflect the driving factors of wetland health in the study area: population density, elevation, and precipitation. Human pressures can disrupt the balance of wetland ecosystems [27]. Population density is a significant threat to wetlands, impacting ecosystem health [28]. Additionally, topographical and climatic factors influence the ecosystem health of wetlands in the study area [29]. Any changes caused by human pressures may disrupt the balance of wetland ecosystems. Population density is one of the threats to wetlands and affects the health of ecosystems. Topographical conditions are a key determinant of wetland spatial distribution, and elevation was chosen to represent the occurrence of severe surface subsidence and changes in surface landscape. Yang et al. [30] argued that precipitation is a major influencing factor for wetland changes, with significant impacts on wetland water resource utilization and vegetation distribution. The driving effect of precipitation on mining subsidence wetlands is reflected by the annual average precipitation.

2.

Pressure subsystem

Environmental pressures caused by human activities are one of the significant factors affecting wetland health [31]. These pressures are primarily derived from regional development, mining subsidence disturbance, road density, and industrial interference. In mining cities with high groundwater levels, human-induced disturbances to wetlands are mainly influenced by land reclamation and expansion of urban construction land [32]. Thus, the regional development index is defined by the ratio of urban construction and agricultural land to the total area of the mining city [33]. The intensity of mining subsidence disturbance significantly affects the spatial distribution of wetlands. Road networks force wetlands to shrink in area and become spatially fragmented [34], and road density is used to represent transportation pressure. Additionally, the higher the proportion of industrial and mining land, the greater the negative impact on wetland ecosystem health.

3.

State subsystem

“State” reflects the wetland ecosystem’s characteristics under the influence of pressures, including its structural and functional features. The Normalized Difference Vegetation Index (NDVI) was used to represent green vegetation cover, with higher values indicating more abundant vegetation [18]. Based on prior studies [35,36,37] and the landscape characteristics of wetlands in mining cities, three indicators—LSI, AI, and SHDI—were selected to characterize the state of subsidence wetlands. The Landscape Shape Index (LSI) quantifies the fragmentation level of the wetland landscape, calculated as the ratio of patch perimeter to total patch area. A higher LSI indicates greater fragmentation. The Aggregation Index (AI) measures the connectivity of similar patches; lower AI values suggest more dispersed, less connected landscapes. The Shannon Diversity Index (SHDI) is commonly used to assess landscape diversity, with higher SHDI values indicating increased patch variety or more even patch distribution, providing a comprehensive measure of landscape heterogeneity.

4.

Impact subsystem

“Impact” refers to the effects of changes in the state of wetland ecosystems on human survival and development. The mining subsidence wetland rate reflects the proportion of mining subsidence wetland area. Hydrological regulation is an important service function of wetland ecosystems, such as flood storage, water supply, and providing habitats and stopover sites for numerous migratory birds [38]. Wetlands have significant infiltration and water storage capacities; they can store precipitation and reduce the volume and delay the timing of runoff entering rivers. The hydrological regulation index is calculated as the ratio of the area of rivers and ditches to the total wetland area.

5.

Response subsystem

“Responses” reflect the management actions taken in reaction to changes in the ecosystem. Under continuous human disturbance, the internal structure and functions of wetland ecosystems are altered, leading to severe landscape fragmentation and a gradual loss of biodiversity [38]. The wetland conservation rate is the ratio of protected wetland area to the total study area, reflecting the intensity of protection efforts. Patch density (PD) represents the degree of patch division and fragmentation in the wetland landscape. Smaller patches and greater fragmentation are indicated by a higher PD value, whereas a lower PD value suggests better preservation of the landscape. In this study, the LSI, AI, SHDI, and PD were calculated by the Fragstats 4.2 software.

2.3.3. Determination of the Assessment Index System

1.

Normalization of Indicator Initial Values

To address differences in the dimensions of various indicators and standardize their values within a consistent range, normalization through positive or negative processing is applied [39]. Assume there are n indicators and m samples. The original data can be represented by the matrix $X={\left(x_{ij}\right)}_{n\times m}$, where $x_{ij}$ denotes the value of the i-th indicator for the j-th sample. After performing normalization, the data are transformed into the standardized matrix $X^{\prime }={\left(x_{ij}^{\prime }\right)}_{n\times m}$, where $x_{ij}^{\prime }$ represents the normalized value corresponding to $x_{ij}$. The classification of indicators as positive or negative depends on their effect on wetland ecosystem health. For positive indicators, higher values signify a more beneficial impact on wetland health. In contrast, for negative indicators, higher values indicate a more detrimental effect on wetland health. The normalization formulas are as follows:

For positive indicators:

$$x_{ij}^{\prime }={\displaystyle \frac{x_{ij}-min{x}_{ij}}{max{x}_i-min{x}_i}}$$

(1)

For negative indicators:

$$x_{ij}^{\prime }={\displaystyle \frac{max{x}_{ij-}{x}_{ij}}{max{x}_i-min{x}_i}}$$

(2)

2.

Measuring Subjective Weights Using the Analytic Hierarchy Process (AHP)

The Analytic Hierarchy Process (AHP) is a systematic traditional method for integrating and analyzing the views of multiple experts [40]. By constructing judgment matrices, calculating matrix eigenvectors, and performing consistency checks, AHP can derive the attribute weights for the evaluation of wetland health status at different levels of factors and indicators.

• Constructing the Judgment Matrix and Calculating the Weight

Based on the field research on coal mining subsidence wetlands, 10 experts were invited to complete the questionnaire and score the indicators. Each expert compared the relative importance of each pair of indicators and scored them according to the 1–9 level judgement matrix scale to determine the final weight of the evaluation indicators. The greater the proportion of the weighted value, the greater the impact of the indicator on the health of the wetland ecosystem.

• Performing Consistency Checks

The following equations were used for the consistency test of the weight values of the assessment indicators:

$$CR={\displaystyle \frac{\mathrm{C}I}{RI}}$$

(3)

$$\mathrm{C}I={\displaystyle \frac{({\mathsf{\lambda }}_{max}-\mathrm{n})}{n-1}}$$

(4)

where ${\mathsf{\lambda }}_{max}$ is the maximum eigenvalue and n denotes the order of the matrix. $RI$ is obtained by checking the consistency test RI value table. It is generally accepted that the data inside the constructed judgement matrix are reasonable and acceptable when $\mathrm{C}I$ < 0.1 and $CR$ < 0.1 are satisfied, otherwise the judgement matrix needs to be reconstructed.

• Determining Subjective Weights

The weight vector is identified as the eigenvector corresponding to the largest eigenvalue, after the consistency check is validated. This vector essentially represents the relative significance of each factor within the system. After this, the weight vector is normalized to ensure that the sum of all weights equals a standardized value. The resulting values are the subjective weights $u_i$ for every indicator.

3.

Measuring Objective Weights Using the CRITIC approach

The CRITIC method is an objective weight allocation approach [41] that has been widely applied in various research fields, including smart communities, stormwater management, and ecological sensitivity assessments [42,43,44]. The correlation coefficients between indicators [45] quantify the conflict, reflecting the degree of disagreement among them [46]. The conflict between indicators is quantified using the correlation coefficients between them, which express the degree of conflict among the indicators. Given the inherent correlations among the indicators of wetland ecosystem health, the CRITIC method is adopted to determine the objective weights. The specific steps of the CRITIC procedure are as follows:

• To calculate the variability (standard deviation $S_i$) of the indicators, we will first calculate the average value of each normalized indicator and then compute the standard deviation. The formulas provided are:

$$\overline{x_i^{\prime }}={\displaystyle \frac{1}{n}}{\sum }_{j=1}^n{x}_{ij}^{\prime }$$

(5)

$$S_i=\sqrt{{\displaystyle \frac{1}{m}}{\sum }_{j=1}^m{\left(x_{ij}^{\prime }-\overline{x_i^{\prime }}\right)}^2} i=1,2,\dots ,n$$

(6)

where $\overline{x_i^{\prime }}$ is the average value of the normalized i-th indicator, $m$ is the total number of samples, and $S_i$ is the standard deviation of the normalized indicator $i$.

• Calculate the conflict between indicators.

The correlation coefficient $r_{ij}$ between two indicators $x_i$ and $x_j$ is calculated as below:

$$r_{ij}={\displaystyle \frac{\sum \left(x_i{\overline{x}}_i\right)\left(x_j{\overline{x}}_j\right)}{\sqrt{\sum {\left(x_i{\overline{x}}_i\right)}^2\sum {\left(x_j{\overline{x}}_j\right)}^2}}}$$

(7)

where $x_i$ and $x_j$ are the scores for the i-th and j-th indicators. ${\overline{x}}_i$ and ${\overline{x}}_j$ are the average scores of the i-th and j-th indicators in wetland health assessment. This correlation coefficient, $r_{ij}$, measures the strength of the relationship between the two indicators. A high positive value indicates a strong positive relationship, while a negative value indicates an inverse relationship.

The conflict between the i-th indicator and all other indicators is calculated by Formula (8):

$$R_i={\sum }_{i=1}^n\left(1-r_{ij}\right)$$

(8)

where $r_{ij}$ is the correlation coefficient between the i-th and j-th indicators. $R_i$ quantifies the overall conflict between the i-th indicator and the others. The higher the value of $R_i$, the greater conflict it has.

• Calculate the information content of the indicators, $G_i$,

$$G_i={\mathrm{S}}_i{\sum }_{j=1}^n\left(1-r_{ij}\right)=S_i{R}_i$$

(9)

where $S_i$ is the standard deviation of the normalized indicator $i$. $R_i$ is the conflict measure for indicator i. A higher $G_i$ indicates more information contained by the i-th indicator, implying that it is more important and should likely have a higher weight.

• Calculate the objective weights of the indicators, $P_i$.

This formula gives the relative importance of each indicator based on its information content.

$$P_i={\displaystyle \frac{G_i}{{\sum }_{j=1}^n{G}_j}}$$

(10)

where $G_i$ is the information content of the i-th indicator. ${\sum }_{j=1}^n{G}_j$ is the sum of information content values for all indicators.

4.

Calculate the combined weights ${\omega }_i$.

The combined weight ${\omega }_i$ for each indicator is calculated using both the AHP weights $u_i$ and the objective weights $P_i$ obtained from the CRITIC method.

$${\omega }_i={\displaystyle \frac{\sqrt{u_i\times p_i}}{{\sum }_{i=1}^n\left(\sqrt{u_i\times p_i}\right)}}$$

(11)

2.3.4. Comprehensive Health Index (CHI) and Classification of Assessment Indicators

1.

Calculation of Criterion Layer Scores, $B_q$.

The normalized score for each criterion layer $B_q$ is calculated by using the weighted sum of the normalized values of the indicators within that criterion layer. This is given by the equation:

$$B_q={\sum }_{i=1}^n{d}_{qi}\times {\omega }_i$$

(12)

where $d_{qi}$ is the normalized value of the i-th indicator that contributes to the q-th criterion layer. ${\omega }_i$ is the weight of the i-th indicator, which reflects its importance in the overall assessment. $B_q$ is the normalized score for the q-th criterion layer.

2.

Calculation of the Comprehensive Health Index (CHI).

After calculating the normalized scores for each criterion layer $B_q$, the Comprehensive Health Index is determined by considering the combined weight of each criterion layer and its normalized score. This is calculated using the equation:

$$\mathrm{CHI}={\sum }_{q=1}^n{\omega }_q\times B_q$$

(13)

where ${\omega }_q$ is the combined weight of the q-th criterion layer. $B_q$ is the normalized score of the q-th criterion layer. The result of this calculation gives the overall health of the wetland, where a higher CHI indicates better wetland health.

$d_{qi}$ is the normalised value of the i-th indictor representing the q-th criterion layer. $B_q$ is the calculated normalised value of the q-th criterion layer, and ${\omega }_i$ is the weight of the i-th indictor.

Referring to previous studies on wetland health assessment criteria [21,28,47,48], in this study we combined the characteristics of the study area, and applied the natural breaks (Jenks) algorithm to establish a five-level classification. The Jenks algorithm aims to minimize within-class variance and maximize the differences between the means of the classes [49]. These classification levels are used as the foundation for assessing the ecosystem health of wetlands in the study area (

Table 3. Indicators for assessing ecosystem health in Yingshang subsidence wetlands.

Level	CHI Value	System Characteristics
Very healthy	0.60–0.83	The wetland ecosystem remains in a favorable natural condition, exhibiting strong vitality and a well-organized structure. External disturbances are minimal, resulting in a highly stable and sustainable system.
Healthy	0.55–0.60	The wetland maintains relatively good ecological integrity, with both its structural components and functional processes largely preserved. Its spatial pattern remains coherent, supporting a stable and enduring ecosystem.
Sub-healthy	0.45–0.55	The wetland ecosystem has been impacted, leading to noticeable structural alterations primarily due to human interference. The system has become highly sensitive to further disturbances.
Poor	0.33–0.45	There is evident degradation within the wetland ecosystem, with signs of deterioration and the partial loss of ecological functions. Efforts to protect and conserve the area are proving insufficient.
Very poor	0.20–0.33	The wetland ecosystem has suffered severe damage, resulting in a critically degraded structure. Extensive fragmentation of wetland patches presents major challenges for effective ecological restoration.

).

3. Results

3.1. Subsystems and Weights of Evaluation Indicators

A 100 m × 100 m fishnet was created for the study area using ArcGIS 10.6. All indicators were intersected with the grid to serve as input variables for the CRITIC model to compute the indicators’ weights. The normalized data values and indicator weight measurement results are shown in

Table 4. Ecosystem health assessment indicators and weights.

Criterion Layer	Weight	Indicator Layer	Combined Weight	Property
Driving Forces (B1)	0.113	Cd1	0.055	Negative
		Cd2	0.005	Negative
		Cd3	0.053	Positive
Pressures (B2)	0.410	Cp1	0.094	Negative
		Cp2	0.278	Negative
		Cp3	0.014	Negative
		Cp4	0.024	Negative
State (B3)	0.286	Cs1	0.045	Positive
		Cs2	0.035	Positive
		Cs3	0.110	Positive
		Cs4	0.096	Positive
Impact (B4)	0.105	Ci1	0.034	Positive
		Ci2	0.071	Positive
Responses (B5)	0.086	Cr1	0.072	Positive
		Cr2	0.014	Negative

. The combined indicator weights reflect their impact on subsidence wetland ecosystem health, as illustrated in Figure 4. The indicators with the greatest impact are the mining subsidence disturbance intensity (CP2), Aggregation Index AI (CS3), and SHDI (CS4). CP2 and CS3 are the two most important indicators, with influence weights of 27.85% and 10.97%, respectively, indicating that coal mining disturbance and AI are key determinants of wetland ecosystem health. Future efforts should focus on wetland protection to reduce the influence of human activities on wetland health. The SHDI (CS4) reflects the status of wetland health, with an influence weight of 9.57%.

The contributions of each subsystem to wetland ecosystem health are shown in Figure 5. The influence weights of the Driving Forces, Pressures, State, Impact, and Response subsystems on wetland ecosystem health are 11.31%, 41.04%, 28.53%, 10.51%, and 8.61%, respectively. Compared with other subsystems, the Pressures and State subsystems have higher influence weights, indicating that these are the primary systems affecting wetland ecosystem health. The Response subsystem holds the lowest weight, indicating that policymakers should prioritize enhancing its significance and allocate more attention to wetland conservation and restoration efforts.

3.2. Analysis of the Comprehensive Health Index (CHI) for Each Subsystem

Three steps were taken to determine the health level of the subsidence wetland. (1) Normalization of indicators: the assessment indicators are normalized according to Equations (1) and (2). (2) Weighting and combined weight calculation: Using both the AHP and CRITIC methods, the normalized values are weighted to derive the subjective and objective weights. The combined weights (CWs) for each indicator are calculated according to Equation (12). (3) Calculation of the Comprehensive Health Index (CHI). A quantitative spatial distribution map of the ecosystem health values for each grid in the study area is generated in ArcGIS 10.6 using the combined weights for calculation.

3.2.1. Driving Forces

The result for Driving Forces is 0.584 (Healthy). Figure 6 illustrates the spatial distribution of each standardized indicator. This indicates that precipitation and topography have minimal disturbance on wetland health, while population density exerts a stronger impact. Jiangkou Town has the highest registered population at 93,168 people, but its administrative area is not large, resulting in the highest population density. Human activities, such as converting wetlands to arable land, over-irrigation, and canalization, have disrupted the wetland ecosystem. The reduction in wetland area, loss of water resources, and degradation of water quality have significantly impacted the habitats, food sources, and shelter of flora and fauna [50].

3.2.2. Pressure

The result for Pressures is 0.690 (Very Healthy). The single indicators from Pressures are shown as Figure 7. The spatial planning, investment, and management by the government or relevant departments have led to good wetland health conditions. However, the regional development index is at the “Sub-healthy” level, mainly due to rapid urbanization. Rivers and wetlands are disturbed by urban construction and land reclamation, leading to increased habitat fragmentation.

3.2.3. State

The score for State is 0.537 (Sub-Healthy). The single indicators from State are shown in Figure 8. The LSI and SHDI have low scores, indicating that the landscape patches are more regular in shape, with lower heterogeneity and fragmentation. Urbanization has changed land use types, with original wetlands being developed or reclaimed, resulting in ecosystem degradation and reduced vegetation diversity.

3.2.4. Impact and Response

Both Impact and Response are at the “Very poor” level, with scores of 0.076 and 0.093, respectively. The single indicators from Impact and Response are shown in Figure 9. The mining subsidence wetland rate, hydrological regulation index, and wetland conservation rate are all at the “Very poor” level. The study area and its surroundings have well-developed water systems with numerous artificial ditches, but their connectivity is poor. The Ji River, which flows west to east through the entire mining area, connects the Liuzhuang Coal Mine subsidence water area and the Xieqiao Coal Mine subsidence wetland, eventually flowing into the Xifei River. However, due to subsidence, the water level has dropped, preventing normal water flow into the ditches, increasing river siltation, and affecting the functionality of water conservancy facilities.

Additionally, Digou Town hosts a national wetland park. The large-scale mining subsidence water area formed by the Xieqiao Mine has significantly impacted this area. In recent years, the park has been approved as a pilot national wetland park by the State Forestry Administration. However, due to the lack of ecological restoration projects, the park relies mainly on artificial interventions and the natural recovery capacity of the ecosystem to improve its current condition. This has led to fragmented patches with no connecting corridors, resulting in ecosystem degradation to some extent.

3.3. Comprehensive Health Index (CHI) of Subsidence Wetland

In 2023, the CHI of the mining subsidence wetlands was 0.517, indicating a sub-healthy condition based on the assessment. The current ecosystem health condition of the mining subsidence wetlands is not optimistic. The main issues include ongoing impacts from mining activities, the fact that the subsidence wetlands have not yet stabilized, significant industrial interference in the surrounding areas, and a relatively weak awareness of wetland protection. To better interpret the spatial pattern of wetland health, a regional statistical analysis was performed, as illustrated in Figure 10.

Very healthy areas comprise 8.86% of the total and are mainly concentrated along the Ji River, linking the mining subsidence wetlands. Additionally, the north of Digou National Wetland Park is included. These regions are characterized by minimal human disturbance, high landscape diversity, and strong wetland protection policies and awareness. Spatially, most of these areas fall within the protection red line of Digou National Wetland Park. These areas can be prioritized for enhanced wetland ecological monitoring and protection.

Healthy areas make up 32.94% of the total, primarily distributed along rivers and ditches, with smaller, scattered patches near the edges of subsidence wetlands in the Xieqiao Mine area. Most of these areas have undergone artificial ecological restoration and management. During the 13th Five-Year Plan period for water conservancy in Yingshang County, key tributaries of the Ji River, including the Xiezhan and Guanghui Rivers, underwent dredging and riparian restoration to enhance ecological function and flood management capacity. In 2021, Xieqiao Mine released the Yingshang County Xieqiao Coal Mine Subsidence Area Lake Protection and Shoreline Protection and Utilization Plan, which zoned the shorelines of mining subsidence wetlands and proposed protection and management measures.

Sub-healthy areas comprise 44.08% of the total, representing the largest proportion within the study area. Wetlands in this area are subject to significant industrial interference. However, they are also protected under the Master Plan for Digou National Wetland Park, with most areas falling within the wetland conservation zone. Wetland protection policies are relatively well-established, and ecological restoration projects are being implemented in an orderly manner.

Areas classified as “Poor” constitute 9.37% of the total study area, reflecting regions with significant ecological degradation and lower wetland health status. They are primarily distributed in the mining subsidence wetlands of Liuzhuang Mine and the central areas of some mining subsidence wetlands in Xieqiao Mine. The mining subsidence wetlands of Liuzhuang Mine are still affected by mining activities, with significant industrial impacts on the wetland landscape and weak protection awareness.

Very poor areas constitute 4.75% of the total, representing the smallest proportion of the wetland area. These areas are sparsely distributed across the southern regions of Liuzhuang Mine and Kouzi East Mine. Wetlands in these areas are heavily impacted by mining and urbanization, with severe wetland fragmentation, increased landscape heterogeneity, and weak awareness and of policies related to wetland protection.

4. Discussion

4.1. Subsidence Wetland Ecological Health Assessment and Indicators

Assessing the health of wetland ecosystems is a highly complex task. To quantitatively evaluate wetland health, various assessment models and indicator systems have been proposed in prior studies [51,52,53]. This study employs the DPSIR model, which provides a practical framework for analyzing the dynamic interplay between the ecosystem health and social development of wetlands in a systematic manner. The indicators constructed within this model are specifically designed for subsidence wetlands in mining areas, drawing on the relevant literature and existing research findings. Since different indicators have varying impacts on wetland ecological health, the determination of indicator weight coefficients directly affects the results. This research used a combined approach, merging the subjective AHP method with the objective CRITIC technique to calculate the weights of the indicators. Das et al. [54] also suggested that combining subjective and objective weighting methods enhances the effectiveness of comprehensive assessments.

In both the AHP and CRITIC methods, mining subsidence disturbance intensity (Cp2) is considered the most influential indicator for wetland ecological health. In Yingshang County, coal mining subsidence wetlands caused by mining activities have become an important component of urban wetlands. As of 2023, the subsidence area formed due to coal mining activities reached as high as 6178.25 hectares. Among them, the subsidence area of Xieqiao Mine reached 2484.62 hectares, and that of Liuzhuang Mine reached 3409.39 hectares. The Liuzhuang Coal Mine features extensive subsidence, rapid expansion, and a high ratio of water accumulation. In contrast, the Xieqiao Coal Mine also has a large subsidence area but expands more slowly, while similarly exhibiting significant water accumulation. This aligns with the findings of He Tingting et al. [55], confirming that the mining subsidence disturbance intensity indicator effectively evaluates the ecological and socio-economic impacts of coal mining subsidence. Conversely, topography is the factor with the least impact on wetland ecological health. In mining cities in the eastern Huaihe Plain, topographical factors have almost no impact on the health of wetlands. Wetland ecosystems are complex and diverse, attributable to the complexity of the environmental system and the differences in the inherent landscape structure within the region. Therefore, indicator selection should be adapted to the specific characteristics of each study area [28].

Wetland ecological health in mining cities is influenced by a wide range of natural, biological, and human activities of various types. These different influencing factors are interwoven and are comprehensively affected by topographical and geomorphological features, altitude, hydrological and meteorological conditions, and human activities [56]. Among these, human activities are the primary drivers affecting wetland ecological health [57]. A comprehensive review of relevant research findings indicates that mining activities, urbanization, and agricultural development are the main forms of human-induced disturbances affecting the landscape evolution of wetlands in the eastern Huaihe Plain [58,59]. Most coal mining subsidence wetlands exist in isolation with low connectivity. Therefore, their ecological functions in regulating rainwater and maintaining biodiversity are relatively weak [60]. Additionally, Acid Mine Drainage (AMD) poses a significant environmental challenge linked to mining activities. AMD pollution causes soil acidification through the release of hydrogen ions, which can severely affect water quality and soil structure, thereby adversely affecting plant growth and nutrient uptake, eventually contaminating the ecosystem [61]. Studies have identified constructed wetlands as the most widely used passive treatment technology for mine water [62]. However, it is necessary to take the acid mine drainage and water quality into consideration while assessing the health condition of the wetland. It could seriously affect the wetland vegetation or wild life habitat. Apart from this, the formation of mining subsidence wetlands has significantly improved habitat diversity and has important potential ecological value. Against this background, balancing wetland socio-ecological health has become an essential component of wetland ecological restoration planning in coal mining cities.

4.2. Planning Strategies of Coal Mining Subsidence Wetlands Based on CHI

Ecological restoration planning is vital for enhancing wetland ecosystem structure and function. However, the instability of coal mining subsidence wetlands means many cities lack adequate landscape-scale assessment and planning [32]. Rational planning of subsidence wetlands is particularly important to avoid secondary disturbances to wetlands during ecological restoration. Based on CHI results, this paper proposes the following strategies and recommendations to further improve wetland ecological health:

(1) The distribution of very healthy wetlands is similar to the spatial distribution of the Digou National Wetland Park and hydrological regulation index (Ci2). Enhancing the monitoring and conservation of wetland ecosystems is essential. In efforts to improve urban environments, many coal mining subsidence wetlands have been transformed into wetland parks through restoration initiatives. There are successful cases of them in the East China region, such as the Pan’an Lake Wetland Park in Xuzhou and the Lvjin Lake Wetland Park in Huaibei [63,64]. The recommendations align with those of Yang et al. [60], noting that while some ecological restoration projects have been successful, the majority of subsidence wetlands still require active development and management. These wetlands hold significant potential for future ecological restoration efforts. National wetland parks, based on protecting and showcasing the wetland ecosystems in plain subsidence areas, further conserve aquatic biological resources and restore the natural water systems and ecological integrity of wetlands, as well as their natural functions.

(2) For healthy wetlands, technical measures such as artificial channels, culverts, and sluices can be used to connect scattered subsidence wetlands, forming an internal interconnected and externally accessible water network. Meanwhile, monitoring, assessment, and early warning should be carried out to enhance wetland biodiversity.

(3) For sub-healthy wetlands, mainly distributed in highly urbanized areas, human activities and industrial disturbances have exacerbated landscape fragmentation. Local planning should prioritize ecological protection and restoration, develop wetland landscape corridors using Nature-based Solutions (NbSs), expand urban public wetland areas, and enhance awareness of wetland ecological protection.

(4) As for poor wetlands, they are still affected by mining activities and should be prioritized for appropriate human intervention to rapidly restore and improve the quality of habitats for plants and animals. For coal mining subsidence wetlands that have stabilized, ecological restoration projects can be further advanced. At the same time, it is necessary to combine local resources and develop agriculture, tourism, and other sustainable industries using coal mining subsidence wetlands.

(5) The very poor areas, disturbed by the Kouzi East and Liuzhuang coal mines, have disrupted the ecological processes of wetlands, leading to the loss and fragmentation of natural habitats. This finding aligns with the reports of Li et al. [65] and Zhang et al. [66], suggesting potential environmental issues such as elevated groundwater levels and widespread waterlogging. It is advised to create ecological buffer zones around the Digou National Wetland Park, enhance ecological source areas and nodes, and improve the structure of the habitat network [67].

4.3. Limitations and Future Directions

This study has some limitations. While the DPSIR model effectively quantifies ecosystem health, it is limited by the selection of indicators. Difficulties in obtaining certain indicators may lead to discrepancies in the assessment results. Future research should focus on conducting deeper investigations into the underlying factors such as species diversity that can directly reflect the ecological functioning and ecological integrity of the subsidence wetland, as well as taking factors into consideration, which may affect the wetland vegetation or wild life habitat by acid mine drainage and water quality. Secondly, the DPSIR framework assumes that environmental change is caused by a single, linear causal relation. In real-world scenarios, however, numerous environmental issues typically arise from the combined effects of multiple interacting factors. These problems may involve complex, non-linear relationships, and it is difficult for purely linear analyses to fully reveal the complexities involved. Despite these limitations, the DPSIR model remains a valuable tool for environmental evaluation and management. Additionally, regular, standardized wetland assessments are essential, and analyzing temporal changes in the CHI can enhance the understanding of ecosystem health across different stages.

5. Conclusions

This study developed a comprehensive ecosystem health assessment framework for coal mining subsidence wetlands, integrating the DPSIR model with AHP and the CRITIC method to provide a novel approach for ecological restoration planning in mining areas. This study provides innovative technical approaches and valuable references for wetland management and planning in Yingshang County. The main conclusions are summarized as follows:

1.

The weights of different indicators and subsystems on wetland ecological health vary. The Pressure and State subsystems have greater impact weights than other subsystems. Key factors affecting wetland ecological health include the regional development index (Cp1), mining subsidence disturbance intensity (Cp2), Aggregation Index (Cs3), Shannon Diversity Index (Cs4), and wetland conservation rate (Cr1).

2.

Analysis of the health status of each subsystem reveals that the scores for “State”, “Impact”, and “Response” are relatively low. In addition, taking into account the spatial distribution of wetland ecosystem health and regional characteristics, strategies for the ecological restoration of coal mining subsidence wetlands have been suggested.

3.

The CHI of coal mining subsidence wetlands is 0.517, indicating a sub-healthy status. Wetlands in sub-healthy, poor, and very poor conditions account for 58.20% of the total wetland area. This is mainly due to the fact that the wetlands in the study region continue to be impacted by ongoing mining operations, and the coal mining subsidence wetlands have yet to reach a state of stabilization. Strengthening ecosystem protection, systematic restoration, and comprehensive management in the region should be considered.