Multi-criteria decision analysis of gas enrichment risk and land degradation in goaf areas

Abstract

Gas enrichment in goaf areas significantly threatens mine safety and surface environmental stability, particularly under top-cutting and pressure-relief mining methods. This study proposes a comprehensive multi-criteria decision-making framework that configures the Analytic Hierarchy Process (AHP) and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) to evaluate a coupled gas enrichment–land degradation risk in goaf areas under top-cutting and pressure-relief mining. Five main criteria, geological conditions, mining parameters, ventilation characteristics, gas emission rates, and ecological vulnerability, were assessed through expert judgment and field data. AHP was used to derive criterion weights, with fault density (11.5%) and coal seam gas content (10.2%) identified as key sub-factors. TOPSIS analysis ranked six goaf zones, identifying G5 as the lowest-risk (C_i = 0.815) and G4 and G6 as the highest-risk areas. Sensitivity analysis confirmed the model’s stability under varied weighting scenarios. The findings reveal a strong link between subsurface gas behavior and surface degradation indicators such as subsidence and vegetation loss. Beyond the generic use of MCDA for multi-factor integration, the study (i) develops a criteria hierarchy that explicitly combines geological, mining, ventilation, and gas-emission drivers with ecological vulnerability indicators, (ii) tailors AHP-derived weights to top-cutting and pressure-relief goaf conditions using a large expert panel, and (iii) quantitatively links the composite risk index to observed land subsidence and vegetation loss to reveal the coupled mechanism between subsurface gas behavior and surface degradation. This model supports early risk detection and informed decision-making, offering practical value for gas control strategies and ecological restoration in mining regions.

Introduction

Coal remains a dominant energy source in many parts of the world, particularly in countries like China, where demand continues to drive deep and extensive underground mining operations^1,2,3. However, new challenges emerge as resource extraction advances into more geologically complex and ecologically sensitive areas^4,5,6. One of the most critical issues is the accumulation of methane and other gases in goaf areas, the collapsed zones left behind after coal extraction^7,8,9,10. If inadequately ventilated or drained, these areas pose severe risks, including gas explosions, spontaneous combustion, and long-term environmental degradation^11,12,13. The increasing adoption of top-cutting and pressure-relief mining techniques, designed to control overburden stress and surface subsidence, has introduced new gas behavior and accumulation complexities, making traditional single-parameter assessments inadequate for modern mining risk management^14,15,16.

Gas enrichment in goaf areas is inherently a multi-factorial phenomenon^17,18,19. It is influenced by a combination of geological structures (e.g., fault density and roof lithology), operational parameters (e.g., mining height and advance rate), ventilation performance, gas emission dynamics, and the ecological sensitivity of the overlying land^20,21,22. Yet, many conventional approaches to risk evaluation in underground mining prioritize technical or environmental factors in isolation, failing to account for their interdependence^23,24,25,26. Moreover, as mining operations extend closer to populated or ecologically valuable zones, the implications of subsurface gas risks increasingly manifest at the surface, causing land subsidence, vegetation loss, and groundwater disturbance^27,28,29. These impacts threaten ecosystems and public safety and complicate post-mining land restoration efforts^30,31,32,33.

Figure 1 provides a conceptual cross-sectional illustration of a goaf area formed after coal extraction under top-cutting and pressure-relief mining methods. The diagram shows key subsurface features, including collapsed roof strata, gas accumulation zones, and ventilation airflow pathways. A drainage borehole is depicted penetrating the overburden to facilitate methane extraction. Airflow is directed through the goaf to dilute and remove gas, reducing explosion and combustion risks. The schematic clearly highlights how geological structure, mining-induced voids, and engineering controls interact to influence gas behavior within the goaf³⁴.

Fig. 1

Conceptual cross-sectional diagram of a goaf area showing gas accumulation, collapsed roof strata, ventilation airflow, and a vertical drainage borehole. This illustration represents key mechanisms influencing methane behavior and control in goaf zones under top-cutting and pressure-relief mining.

Effective risk assessment in underground mining environments has gained increasing attention in recent years due to the hazardous nature of subsurface operations^35,36,37. Various decision-making frameworks have been applied to mitigate threats such as gas explosions and environmental pollution^26,38,39. A study by⁴⁰ developed a hybrid fuzzy AHP-TOPSIS model to evaluate and select optimal methane explosion control methods in underground coal mines. The methodology involved identifying 34 sub-risk factors, with expert surveys used to collect data and assess hazard weights. “Improper ventilation system” emerged as the highest-ranked risk factor. Using fuzzy TOPSIS, different control strategies were evaluated, and the results emphasized that investments in safety technology and financial infrastructure are the most effective preventive approaches. The study also included a sensitivity analysis, confirming the robustness of the rankings. This research demonstrated the utility of fuzzy MCDM tools for dynamic and multi-factorial hazard evaluation in high-risk mining environments. Another study by⁴¹ addressed the environmental risks associated with underground reservoirs storing concentrated saltwater. A risk evaluation model was proposed to assess the pollution potential under both leakage and non-leakage scenarios. It was found that the complexity of mine water quality had the most significant impact on risk grading. Applied to the Lingxin Mining Area, the analysis showed a 60% probability of low environmental risk under non-leakage conditions; however, the risk level escalated dramatically when saltwater leakage was introduced. This research highlights the importance of uncertainty modeling and the role of hydrogeological parameters in underground risk management. Both studies demonstrate the versatility and reliability of multi-criteria decision-making frameworks (especially AHP and TOPSIS variants) for handling complex mining risks. The first study emphasizes safety-related explosion hazards, while the second focuses on environmental pollution from underground storage systems. However, existing MCDM applications typically treat mine safety and environmental impacts as separate decision problems and rarely target gas enrichment in goaf areas formed by top-cutting and pressure-relief mining. In this context, the present study advances the MCDM-based risk assessment paradigm in three ways. First, it formulates a coupled concept of gas enrichment–land degradation risk. It builds a criteria hierarchy that explicitly links geological structure, mining parameters, ventilation, and gas-emission behavior with surface ecological vulnerability indicators. Second, it configures and calibrates an AHP–TOPSIS model specifically for goaf areas under top-cutting and pressure-relief mining, using expert-derived weights and sensitivity analysis to identify the dominant risk drivers. Third, it demonstrates how the resulting composite risk index can be spatially mapped and statistically related to subsidence and vegetation loss, thereby revealing the joint mechanism between subsurface gas dynamics and land degradation. While previous works focus on discrete risk types or accident control, this study therefore provides a holistic yet mechanism-oriented evaluation of gas enrichment risk that integrates underground hazards with surface environmental responses.

The present study introduces a novel decision-support framework to evaluate gas-enrichment risk in goaf areas affected by top-cutting and pressure-relief mining techniques, which are increasingly used in modern deep coal mining operations. Unlike traditional risk assessments focusing narrowly on gas emission rates or isolated geological indicators, this study integrates a broader set of criteria encompassing geological conditions, mining parameters, ventilation systems, gas emission characteristics, and ecological vulnerability. This multidimensional approach enables a more holistic understanding of how subsurface gas behavior interacts with surface land-degradation processes, an area that has received limited attention in previous studies. The innovation of this research lies not in the use of AHP and TOPSIS per se, but in how they are combined and parameterized to capture the coupled gas enrichment–land degradation system in goaf areas under top-cutting and pressure-relief mining, providing a systematic, transparent, and scalable basis for ranking zones according to this composite risk. While AHP enables quantifying expert judgment across multiple disciplines, TOPSIS transforms these judgments into actionable priority rankings. This dual-method framework enhances the robustness of both the qualitative and quantitative components of the evaluation. Moreover, conducting a sensitivity analysis adds an extra layer of credibility by examining the stability of rankings across various weighting scenarios, thereby ensuring the model’s reliability in real-world conditions. In summary, this study contributes to the literature in three main aspects. (1) Conceptually, it introduces a coupled gas enrichment–land degradation risk framework for goaf areas affected by top-cutting and pressure-relief mining, bridging mine safety and land rehabilitation perspectives. (2) Methodologically, it develops a tailored AHP–TOPSIS model with a new indicator system and expert-derived weights for this specific mining context and tests its robustness through multiple weighting scenarios. (3) Empirically, it applies the framework to six representative coal mines in China to identify dominant risk factors (such as fault density and coal seam gas content) and to demonstrate a quantitative association between the composite risk index and observed surface degradation indicators.

The study is structured as a sequence of four connected stages that move from context to application and synthesis. It begins with a problem and literature stage, where the background of gas enrichment in goaf areas, its links to land degradation, and gaps in existing risk assessment approaches are clarified. The second stage develops the methodological framework, configuring the AHP–TOPSIS model, defining the criteria system, and incorporating expert judgement and sensitivity analysis to test the robustness of the weighting scheme. The third stage applies this framework to six representative goaf zones, combining monitoring and geological data to compute risk indices, map spatial patterns of gas enrichment, and interpret the results considering both subsurface processes and surface ecological conditions. The final stage synthesizes the main findings, highlights methodological and practical implications for gas control and land rehabilitation, acknowledges limitations, and proposes directions for future research on sustainable mine management and land degradation mitigation (Fig. 2).

Fig. 2

Structural flow of the study, from the problem context and literature review through the methodological framework and case-study application to the synthesis of implications for mining risk management and land degradation control.

Methodology

To evaluate gas enrichment risks in goaf areas under top-cutting and pressure-relief mining, a Multi-Criteria Decision-Making (MCDM) framework was constructed by integrating the Analytic Hierarchy Process (AHP) with the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). The AHP was employed to determine the weights of evaluation criteria based on expert judgment, while TOPSIS was used to rank various risk scenarios and mitigation strategies. This section details the steps in model construction, data collection, and evaluation. The framework for evaluating gas enrichment risk in goaf areas under top-cutting and pressure-relief mining integrates multi-criteria evaluation with expert-based weighting and quantitative ranking, combining the strengths of AHP (Analytic Hierarchy Process) and TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution). This hybrid framework enables a structured analysis of multiple interrelated criteria influencing gas enrichment risk and supports informed decision-making for land degradation control. Gas migration in goaf areas is inherently nonlinear and time-dependent, especially during roof collapse, goaf compaction, and the progressive development of fractures. In this study, the AHP–TOPSIS framework is therefore formulated as a quasi-steady risk zoning tool: time-varying processes are represented indirectly through indicators that summarize their cumulative or typical effects over the relevant operating period rather than by explicitly simulating transient gas flow. The aim is not to replace detailed numerical simulations at the face scale, but to provide a transparent, data-efficient decision-support scheme for comparing the overall risk levels of multiple goaf zones under top-cutting and pressure-relief mining. The framework consists of the following sequential steps:

• I. Hierarchical structuring of the evaluation problem:

  A three-level hierarchical structure was developed to organize the decision problem:

  • Level 1: Goal

  Evaluation of gas enrichment risk in goaf areas under top-cutting and pressure-relief mining.

  • Level 2: Criteria

    • C1: Geological Conditions

    • C2: Mining Parameters

    • C3: Ventilation Characteristics

    • C4: Gas Emission Rates

    • C5: Ecological Vulnerability

  • Level 3: Sub-Criteria

    Each criterion was broken down into measurable sub-criteria (to be detailed in a table later).

    • C1: Geological Conditions includes fault density, coal seam gas content, and roof lithology.

    • C2: Mining Parameters includes mining depth, top-cutting range, and face length.

    • C3: Ventilation Characteristics includes air volume, pressure gradient, and airflow stability.

    • C4: Gas Emission Rates includes measured methane flux, residual gas pressure, and drainage efficiency.

    • C5: Ecological Vulnerability includes vegetation cover index, land use sensitivity, and distance to water bodies (Fig. 3).

Fig. 3

The hierarchical structure of the evaluation framework shows the main criteria and their corresponding sub-criteria used to assess gas enrichment risk in goaf areas under top-cutting and pressure-relief mining.

The selection of the 15 secondary criteria was based on a structured process that combined a literature review, regulatory requirements, data availability, and expert judgement. First, an initial long list of potential indicators was compiled from previous studies on gas enrichment and outburst risk, mining-induced subsidence and land degradation, and ecological impacts in coal mining areas, as well as relevant national technical standards and mine safety guidelines. Second, this list was screened according to three filters: (i) conceptual relevance to the coupled gas enrichment–land degradation framework, (ii) measurability and data consistency across all six case-study mines, and (iii) avoidance of apparent redundancy among indicators describing the exact physical mechanism. Third, the remaining indicators were refined into the final set of 15 sub-criteria through a structured expert workshop, during which domain experts grouped and merged overlapping items and ensured that key geological, operational, ventilation, gas-emission, and ecological processes were all represented. Given the relatively small number of goaf zones and the heterogeneous data sources, formal correlation analysis or regression-based feature selection was not applied at this stage, as it would be statistically unstable and could lead to spurious exclusion of physically important factors.

• II. Weight determination using AHP

  The AHP method was applied to derive weights for both main criteria and sub-criteria. This involved expert-based pairwise comparisons to reflect the relative importance of each factor in contributing to gas enrichment risk. A consistency check (via Consistency Ratio) ensured the reliability of the judgments. Criteria with CR > 0.1 were re-evaluated to maintain decision coherence.

• III. Construction of decision matrix:

  A decision matrix was constructed based on field surveys and monitoring data. Each row represents a distinct mining zone or scenario, while columns correspond to sub-criteria values. All data were normalized to make them dimensionless and comparable. Normalization increased with magnitude for benefit-type criteria (where higher values are more favorable). Normalization decreased with magnitude for cost-type criteria (where higher values indicate risk).

• IV. TOPSIS-based risk scoring:

  TOPSIS was then used to score and rank each alternative (mining scenario) based on its proximity to the ideal solution:

  • An ideal (best-case) and anti-ideal (worst-case) scenario were established.

  • The normalized, weighted matrix from AHP was used to calculate Euclidean distances to both ideal and anti-ideal points.

  • The relative closeness of each alternative to the ideal solution was computed, providing a quantifiable risk score.

• V. Interpretation and decision support

  The final TOPSIS scores enabled prioritizing high-risk areas and selecting suitable gas control and ecological mitigation strategies. Visual risk zoning maps were produced using GIS tools to link gas risk with surface degradation patterns spatially. These outputs support early warning systems, guide mine planning, and inform targeted land remediation efforts.

Analytic hierarchy process (AHP) methodology

The Analytic Hierarchy Process (AHP) is a structured technique for organizing and analyzing complex decisions based on mathematics and psychology. In this study, AHP was applied to derive the relative weights of the criteria and sub-criteria influencing gas enrichment risk in goaf areas. The AHP process was conducted as follows:

• a. Pairwise comparison matrix construction

  A pairwise comparison matrix was developed for both the main criteria and sub-criteria. Each expert compared two elements at a time using the Saaty scale, which ranges from 1 (equal importance) to 9 (extreme importance). For n criteria, an n × n matrix A = [a_ij] is formed, where:

  • a_ij is the importance of criterion i over criterion j

  • a_ji = 1/a_ij

• b. Normalization of the matrix

  Each element of the matrix is divided by the sum of its column to obtain a normalized matrix N = [n_ij]:

$$n_{ij} = \frac{{a_{ij} }}{{\sum\nolimits_{i = 1}^{n} {a_{ij} } }}$$

• c. Computation of the priority vector (Weights)

  The weights of the criteria (priority vector w_i) are obtained by averaging the values across each row of the normalized matrix:

$$w_{i} = \frac{1}{n}\left( {\sum\limits_{j = 1}^{n} {n_{ij} } } \right)$$

• d. Consistency check

  To ensure consistency in the pairwise comparisons, the following steps are performed:

  • Compute the weighted sum vector AW by multiplying the pairwise matrix A with the priority vector W

  • Compute the Consistency Index (CI):

$$\lambda_{max} = \frac{{\left( {AW} \right)_{i} }}{{\left( W \right)_{i} }} , CI = \frac{{\lambda_{max} - n}}{n - 1}$$

• Compute the Consistency Ratio (CR):

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

RI is the Random Index, a benchmark value dependent on the size of the matrix. Matrices with a CR ≤ 0.10 were considered acceptably consistent. In cases where the CR exceeded this threshold, follow-up consultations were conducted with the corresponding respondents to resolve inconsistencies. Only after all matrices passed the consistency test where they finalized.

• e. Expert integration and final weights

  To ensure the robustness and credibility of the AHP-derived weightings, a panel of 30 subject matter experts was carefully selected. Henan University confirmed that all questionnaires and the online protocol were completed, and the obtained data are confidentially reserved to maintain the anonymity of the surveys. All methods were carried out in accordance with relevant guidelines and regulations. The study was approved by the Ethics Committee of Henan University, China. These individuals were drawn from academia and industry, representing diverse expertise in mining engineering, rock mechanics, ventilation systems, gas control, and environmental management^42,43. This balanced composition was intended to capture comprehensive perspectives on gas enrichment risk and its ecological implications. The panel included:

  • 12 academic experts (40%) from leading universities and research institutes, specializing in underground mining technologies, geological hazard assessment, and land degradation modeling.

  • 18 industry professionals (60%), including senior mine safety engineers, ventilation officers, gas drainage specialists, and environmental restoration consultants from major mining enterprises and regulatory bodies.

All selected experts possessed at least 10 years of experience in their respective fields, with several having participated in national-level mining safety audits or environmental impact assessments. This ensured that both theoretical frameworks and on-ground operational realities were adequately represented. A structured AHP questionnaire was developed based on the hierarchical model outlined in the framework design. The questionnaire was divided into two main parts:

• 1. Pairwise comparisons of the five primary criteria influencing gas enrichment risk (i.e., geological conditions, mining parameters, ventilation characteristics, gas emission rates, and ecological vulnerability).

• 2. Sub-criteria comparisons within each main criterion category (e.g., within “geological conditions”: fault density, coal seam gas content, roof lithology).

Each pair was rated using the standard Saaty 1–9 importance scale, where a score of 1 indicates equal importance and 9 indicates extreme importance of one factor over another. The questionnaire was formatted using an online platform, supplemented with downloadable forms for participants preferring offline responses. A pilot test was conducted with three experts before full distribution to ensure clarity. Based on their feedback, minor revisions were made to improve question structure and eliminate ambiguity. The finalized questionnaire was then distributed via email to all 30 experts. Participants were given two weeks to respond, with follow-up reminders to maximize participation. A total of 26 valid responses were received (response rate: 87%). The geometric mean method was employed to synthesize the multiple expert judgments into a single set of pairwise comparison matrices. This method is well-suited for aggregating AHP inputs, as it preserves the original data’s multiplicative properties and ratio scale.

For any two elements i and j, the aggregated value ${a}_{ij}^{*}$ is computed as:

$$a_{ij}^{*} = \left( {\prod\limits_{k = 1}^{m} {a_{ij}^{\left( k \right)} } } \right)^{{{\raise0.7ex\hbox{$1$} \!\mathord{\left/ {\vphantom {1 m}}\right.\kern-0pt} \!\lower0.7ex\hbox{$m$}}}}$$

where:

• ${a}_{ij}^{(k)}$ is the pairwise comparison value from the kth expert,

• m is the number of experts (in this case, 26).

This process generated consolidated comparison matrices for the main criteria and each group of sub-criteria. Finally, the normalized weights for each criterion and sub-criterion were derived from the adjusted matrices. These weights reflect the relative influence of each factor in contributing to gas enrichment risk and serve as inputs for the subsequent TOPSIS-based risk evaluation.

TOPSIS methodology

The Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) is a widely adopted multi-criteria decision-making method that ranks alternatives based on their relative distance from an ideal solution. This study applied TOPSIS after determining the criterion weights through AHP to evaluate and rank the gas enrichment risk across different goaf areas. The method integrates the positive (ideal) and negative (non-ideal) aspects of decision outcomes, offering a clear and rational basis for prioritization. The implementation of TOPSIS in this study involved the following structured steps:

• a) Construction of the decision matrix

  A decision matrix D = [x_ij] was constructed, where:

  • i = 1,2,…,m represents the alternatives (i.e., different goaf zones or mining scenarios),

  • j = 1,2,…,n represents the evaluation sub-criteria.

  Each element x_ij represents the performance score of alternative i concerning sub-criterion j, obtained from field measurements, mine monitoring systems, and ecological assessments. For qualitative sub-criteria, specifically Roof Lithology (C1.3), Intake and Return Configuration (C3.2), Sealing Efficiency (C3.3), and Land Use Sensitivity (C5.3), categorical information was converted into ordinal risk scores on a 1–5 scale using explicit grading rules. All scales were defined so that higher scores represent a greater contribution to gas enrichment or land-degradation risk, and these sub-criteria were treated as cost-type indicators in the normalization step. A detailed scoring manual was prepared and provided to the evaluators to ensure consistent use of the categories across mines. To reduce individual subjectivity and improve repeatability, each qualitative sub-criterion for every goaf zone was scored independently by at least three experts familiar with the local geological, ventilation, and land-use conditions. The final score used in the decision matrix is the median of these ratings. When the difference between the highest and lowest rating for a given item exceeded one grade level, the experts jointly reviewed the underlying information. They agreed on a consensus category before finalizing the score.

• b) Normalization of the decision matrix

  To eliminate the effects of differing units and scales, the raw data matrix was normalized using vector normalization:

$$r_{ij} = \frac{{x_{ij} }}{{\sum\nolimits_{i = 1}^{m} {x_{ij}^{2} } }}$$

This transformation yields the normalized decision matrix R = [${r}_{ij}$].

• c) Weighted normalized decision matrix

  The weights w_j derived from AHP were applied to each normalized criterion to form the weighted normalized matrix V = [v_ij]:

$$v_{ij} = {\text{w}}_{{\text{j}}} .r_{ij}$$

This step ensures that the importance of each criterion is incorporated into the decision-making process.

• d) Determination of ideal and non-ideal solutions

  The positive ideal solution A⁺ and negative ideal solution A⁻ were determined as follows:

  $A^{ + } = \left\{ {\max \left( {v_{ij} } \right)for\,benefit\,criteria;\,\min \left( {v_{ij} } \right)for\,\cos t\,criteria} \right\}$;

  $A^{ - } = \left\{ {\min \left( {v_{ij} } \right)for\,benefit\,criteria;\,\max \left( {v_{ij} } \right)for\,cost\,criteria} \right\}$;

  These represent the best and worst possible performances for each criterion.

• e) Calculation of separation measures

  The Euclidean distance of each alternative from the ideal and non-ideal solutions was calculated:

$$S_{i}^{ + } = \sqrt {\mathop \sum \limits_{j = 1}^{n} \left( {\nu_{ij} - v_{j}^{ + } } \right)^{2} }$$

$$S_{i}^{ - } = \sqrt {\mathop \sum \limits_{j = 1}^{n} \left( {\nu_{ij} - v_{j}^{ - } } \right)^{2} }$$

where:

• ${S}_{i}^{+}$ is the distance from the ideal solution,

• ${S}_{i}^{-}$ is the distance from the non-ideal solution.

• f) Calculation of relative closeness

  The relative closeness to the ideal solution for each alternative is then computed as:

$$C_{i} = \frac{{S_{i}^{ - } }}{{S_{i}^{ + } + S_{i}^{ - } }}$$

The value of ${C}_{i}$ ranges between 0 and 1:

• A value closer to 1 indicates higher similarity to the ideal solution (i.e., lower gas enrichment risk),

• A value closer to 0 indicates higher similarity to the worst-case scenario (i.e., higher risk).

• g) Ranking and risk classification

  The computed ${C}_{i}$ values ranked alternatives (i.e., goaf zones) from highest to lowest. Thresholds were defined to classify each zone into risk categories:

  • Low Risk: ${C}_{i}$> 0.75

  • Moderate Risk: 0.50 < ${C}_{i}$≤ 0.75

  • High Risk: Ci ≤ 0.50

These classifications were later visualized through risk zoning maps using GIS tools to guide mitigation efforts and land restoration planning.

Definition of criteria and sub-criteria

Evaluating the risk of gas enrichment in goaf areas under top-cutting and pressure-relief mining requires a multidimensional assessment framework. Five primary criteria were identified based on an extensive academic literature review, industry guidelines, and expert consultations. Each criterion captures a specific dimension of risk related to gas accumulation and its environmental consequences, particularly land degradation. Although many underlying processes (such as gas emission, fracture propagation, and subsidence) evolve in time and exhibit nonlinear behavior, they are incorporated here through indicators that aggregate their effects over characteristic periods of the top-cutting and pressure-relief stage. Examples include long-term average gas emission rates, observed fracture development intensity, and accumulated subsidence or vegetation changes. This approach allows dynamic phenomena to be reflected within a static multi-criteria representation that can be consistently populated and compared across all investigated mines. The criteria were subdivided into measurable sub-criteria to enhance resolution and facilitate data collection. Each criterion and its associated sub-criteria are briefly described in Table 1. These were selected based on the following considerations:

• Relevance: Factors must directly or indirectly influence gas migration, accumulation, or dissipation in goaf areas.

• Data Availability: Sub-criteria had to be quantifiable through field monitoring, geotechnical data, or expert estimation.

• Ecological Significance: Sub-criteria with implications for land degradation (e.g., subsidence, soil pollution, or vegetation loss) were prioritized.

• Expert Consensus: The final set was refined based on feedback from the 30-member expert panel involved in the AHP stage, ensuring the framework reflects both academic rigor and operational practicality.

Table 1 Criteria and sub-criteria used in the multi-criteria decision analysis (MCDA) framework for evaluating gas enrichment risk in goaf areas.

From a risk-mechanism perspective, ecological vulnerability in coal mining subsidence areas is closely associated with the intensity of mining-induced deformation. Remote-sensing and field studies have shown that zones with strong subsidence and ground fissuring tend to exhibit severe vegetation degradation, conversion to bare land, and loss of ecological function⁴⁴. These deformation-related surface cracks and subsidence depressions provide preferential pathways and outlets for coal mine gases migrating from the goaf to the surface and can locally favour the accumulation of hazardous gas concentrations in enclosed or topographically low-lying areas⁴⁵. In the proposed framework, ecological indicators therefore play a dual role: they reflect (i) the potential severity of surface ecological and socio-economic consequences and (ii) the degree to which the surface environment has been modified by subsidence and fissuring in ways that affect gas dissipation and accumulation conditions above the goaf.

Monitoring data and measurement protocols

Several sub-criteria in the indicator system are derived from on-site monitoring data. To ensure data reliability and traceability, all measurements used in this study follow the routine monitoring protocols of the participating mines. Gas concentration in roadways and key goaf-adjacent locations is measured using fixed, continuously recording gas sensors (logging interval 1–5 min, accuracy better than ± 0.5% CH₄ volume fraction), calibrated according to the manufacturer’s specifications and cross-checked with portable gas detectors during regular inspections. Gas emission and drainage rates are obtained from flowmeters installed on drainage pipelines, with readings automatically logged at least once per shift and manually verified daily by mining staff. Ventilation parameters (airflow, air pressure) are determined from anemometer and manometer measurements conducted by ventilation technicians at least monthly and after significant ventilation adjustments. Groundwater depth is taken from hydrogeological boreholes and mine water-level monitoring points, which are typically checked monthly or after major hydrological events. Subsidence data are compiled from the mine’s routine surface deformation surveys and levelling campaigns, which are conducted at intervals of several months to one year, depending on the stage of mining. Ecological indicators, such as vegetation coverage, are derived from interpreted orthophotos and satellite imagery with known spatial resolutions, complemented by field observations for verification. Geological structure, fault density, and roof lithology are obtained from exploration borehole logs and geological maps and are therefore not subject to instrumental precision issues but to the underlying quality of the geological investigation (see Table 2).

Table 2 Monitoring data sources, frequency, and instrument precision for sub-criteria based on monitoring.

Ecological vulnerability was quantified using spatially explicit data rather than purely qualitative descriptions. For each goaf zone, Vegetation Coverage (C5.1) was expressed as the proportion of the surface area classified as vegetated within the delineated influence area, derived from the interpretation of up-to-date orthophotos and satellite imagery, combined with field observations of vegetation condition. The resulting coverage values were grouped into ordered classes (sparse to dense) and converted to standardized scores for use in the decision matrix. Groundwater Depth (C5.2) was obtained from hydrogeological borehole records and mine water-level monitoring reports and then averaged over the footprint of each goaf zone. Land Use Sensitivity (C5.3) was determined from official land-use maps and mine planning documents by assigning each land-use category (e.g., industrial land, cropland, residential areas, ecological protection areas) to a sensitivity level based on its ecological and socio-economic importance. These procedures ensured that ecological indicators were based on reproducible spatial datasets, with field surveys used primarily for verification rather than as the sole source of information. For qualitative sub-criteria, specifically Roof Lithology (C1.3), Intake and Return Configuration (C3.2), Sealing Efficiency (C3.3), and Land Use Sensitivity (C5.3), categorical information was converted into ordinal risk scores on a 1–5 scale using explicit grading rules, as summarized in Table 3, where higher scores represent a greater contribution to gas enrichment or land-degradation risk.

Table 3 Grading rules for qualitative sub-criteria used in the AHP–TOPSIS model.

Results

The integrated application of the Analytic Hierarchy Process (AHP) and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) enabled a structured, quantitative evaluation of gas-enrichment risk in goaf areas under top-cutting and pressure-relief mining methods. The methodology was applied to a representative mining region experiencing severe land degradation, frequent surface subsidence, ground fissures, and vegetation loss. By incorporating expert judgment, geological and environmental data, and operational parameters into a hierarchical model, the study systematically identified the most critical risk factors. It evaluated the severity of gas enrichment across multiple sites. The results provide valuable insights into the interaction between underground gas dynamics and surface environmental impacts, revealing the spatial variability of gas risk and its correlation with land degradation indicators. This section presents the main findings in a structured format, beginning with the computed weights of criteria and sub-criteria using AHP, followed by the TOPSIS-based risk ranking of goaf zones, and finally, a synthesis of the implications for land degradation control and risk mitigation planning.

AHP-derived weights of criteria and sub-criteria

Following expert input aggregation, the final priority weights for the main criteria and their corresponding sub-criteria were computed. These weights reflect the relative importance of each factor in contributing to gas enrichment risk. Table 4 presents the normalized weights derived from the geometric mean of expert judgments, with the consistency ratios (CR) within acceptable thresholds (CR < 0.1 for all matrices).

Table 4 AHP-derived normalized weights for main criteria and sub-criteria influencing gas enrichment risk in goaf areas.

Table 4 shows that Geological Conditions (26.6%) and Mining Parameters (21.4%) were considered the most influential categories in the overall gas enrichment risk assessment framework. This emphasizes that the intrinsic geological structure and the technical execution of mining operations are the primary drivers of gas behavior in goaf areas. Within the Geological Conditions criterion, Fault Density (C1.1) received the highest sub-criterion weight (0.115), highlighting its critical influence on gas migration and accumulation. Faults act as conduits for methane and other gases, enabling rapid upward or lateral flow into goaf spaces where gas may accumulate if not adequately ventilated or drained. Moreover, dense fault networks often correlate with fractured and unstable strata, increasing the risk of gas escape and subsequent ground deformation, such as surface subsidence and cracking.

Coal Seam Gas Content (C1.2), with a sub-weight of 0.102, was also ranked highly, underscoring that the initial gas potential within the coal seam is a key determinant of enrichment risk. A high methane content means a greater volume of gas is released upon mining disturbance, which, if inadequately managed, can lead to hazardous concentrations in enclosed goaf spaces. Roof Lithology (C1.3), though weighted lower (0.049), still influences fracture propagation. Soft or porous roof strata may absorb or allow gas transmission more readily, whereas hard, impermeable roofs may trap gas below, increasing localized pressure. Mining Parameters, the second-highest category (21.4%), includes factors that are largely operational but deeply intertwined with risk development. Mining Height (C2.1) (0.089) affects the void size left after extraction; a greater height results in larger voids and more extensive caving, promoting greater fracture development and potentially larger gas reservoirs in the overburden. Advance Rate (C2.2) (0.078) influences the stress redistribution around the face; faster rates often result in more abrupt rock failure and uncontrolled gas release, while slower advancement may facilitate more stable drainage. Top-Cut Thickness (C2.3) (0.047), though less influential, modifies the caving characteristics of the immediate roof, affecting the compaction of broken rocks and the permeability of the goaf.

Gas Emission Rates, with a total weight of 19.9%, also played a significant role in the model. Emission from Coal Seam (C4.1) (0.082) stood out among all gas-related sub-criteria, further reinforcing that gas release at the face is a primary concern. Residual Gas Emission (C4.2) (0.069) considers gas from fragmented coal and surrounding rock, which can persist in the goaf long after mining has progressed. Delay in Gas Drainage (C4.3) (0.048) reflects the time lag between mining disturbance and the implementation of gas control measures; longer delays can allow a significant buildup of methane and increase the potential for spontaneous combustion. Ventilation Characteristics (18.5%) accounted for key engineering controls that mitigate or exacerbate risk. Airflow Velocity (C3.1) (0.071) ensures sufficient gas dilution; insufficient airflow may allow localized accumulations to exceed explosive limits. The Intake and Return Configuration (C3.2) (0.066) determines how air circulates and whether it effectively reaches the goaf or bypasses critical areas. Sealing Efficiency (C3.3) (0.048) is crucial to isolating abandoned or non-ventilated voids; weak seals may leak gas into active workings or allow air ingress that supports spontaneous combustion.

Finally, Ecological Vulnerability (13.6%) had the lowest total weight among the main criteria, reflecting its more indirect influence on gas enrichment dynamics compared with geological and operational drivers. Vegetation Coverage (C5.1) (0.062) provides a proxy for surface ecological health and is often negatively affected by underlying gas pressures and land instability. Groundwater Depth (C5.2) (0.045) is important because shallow water tables may influence gas migration and oxidation pathways, and changes in water levels may also indicate subsurface void formation. Land Use Sensitivity (C5.3), the lowest-weighted sub-criterion (0.029), refers to the socio-economic and ecological importance of the surface land. Although it does not directly control gas behavior underground, it is critical for assessing the consequences of gas-related hazards (explosions, sinkholes, contamination) on sensitive or inhabited areas. The resulting weight pattern reflects the expert panel’s collective judgment under current safety regulations, where preventing gas accidents is treated as the primary objective, and ecological impacts, while important, are considered a secondary consequence. Notably, the sensitivity analysis explicitly includes a “High Ecological Sensitivity” scenario in which the weight of ecological vulnerability is increased by 30%; in that case, zones with higher ecological exposure (e.g., G2 and G6) move upward in priority, demonstrating that the framework can be readily recalibrated to align with stricter environmental policies or sustainable development goals. In summary, geological and mining-induced factors dominate the occurrence probability of gas enrichment events, while ecological indicators shape the assessment of their potential consequences. Together, they form a weighting scheme that balances technical accuracy with ecological concern and remains flexible enough to support alternative policy priorities through reweighting.

TOPSIS-based risk evaluation of goaf areas

The computed AHP weights were applied in the TOPSIS framework to evaluate and rank six goaf areas (labeled G1-G6) based on their relative gas-enrichment risk. Field data for each sub-criterion were collected through an integrated approach involving borehole gas pressure monitoring, drainage system performance logs, and ecological field surveys assessing surface vegetation degradation and subsidence. After preprocessing, all data were normalized using vector normalization methods and weighted according to the final AHP-derived sub-criterion importance values. The weighted decision matrix was then used to compute the Euclidean distances of each goaf area from the positive ideal solution (representing optimal safety and minimal gas risk) and the negative ideal solution (representing worst-case enrichment scenarios). The relative closeness values _Ci, calculated as the ratio of the distance to the anti-ideal over the sum of both distances, quantify each area’s proximity to the desired condition, with higher values indicating lower risk. These C_i values were used to classify the goaf zones into high, moderate, or low gas enrichment risk levels, directly informing the prioritization of safety and environmental mitigation strategies. The six goaf zones used in this study were selected from representative coal mines in China’s geological and ecological regions, each known to apply top-cutting and pressure-relief mining methods. These include:

• G1: Jincheng Western Mine, Shanxi Province

• G2: Pingshuo East Open-Pit Transition Zone, Shanxi Province

• G3: Huainan Panji Mine, Anhui Province

• G4: Pingdingshan No. 8 Mine, Henan Province

• G5: Tongchuan Yuhua Mine, Shaanxi Province

• G6: Tiefa Mine, Liaoning Province

These sites were chosen to reflect diverse geological structures, ventilation systems, gas content profiles, and degrees of ecological vulnerability, enabling a comprehensive evaluation of gas enrichment risk under varied mining conditions within China’s major coal-producing regions. Table 5 shows the final relative closeness values, Ci, for each goaf zone, representing its proximity to the optimal condition (low risk). These values were then used to classify each area into a risk level. Figure 4 presents the spatial distribution of gas enrichment risk across six representative goaf zones in China, as assessed using the AHP-TOPSIS model. The map uses a color-coded classification scheme, green for low risk, yellow/orange for moderate risk, and red for high risk, to indicate the relative hazard levels of each site. Notably, G5 (Tongchuan Yuhua Mine) consistently ranks as the lowest-risk area due to favorable geological and operational conditions, whereas G4 (Pingdingshan No. 8 Mine) is classified as high risk, reflecting its dense fault structure and ineffective drainage. It provides a geospatial visualization that complements the analytical results, supporting targeted mitigation planning and region-specific ecological restoration efforts.

Table 5 TOPSIS evaluation of goaf areas based on gas enrichment risk.

Fig. 4

Spatial distribution of gas enrichment risk in goaf areas across six mining zones in China based on AHP-TOPSIS evaluation. Risk levels are color-coded: green (low), yellow/orange (moderate), and red (high), with G5 as the safest and G4 as the most hazardous. These visual supports spatial prioritization for gas control and land degradation mitigation.

The TOPSIS analysis results, derived from the weighted AHP framework, revealed notable variations in gas enrichment risk across the six evaluated goaf areas. Among them, G5 (Tongchuan Yuhua Mine) emerged as the least risky zone, exhibiting the highest relative closeness value (C_i) to the positive ideal solution. This ranking reflects a combination of favorable geological and operational factors. Specifically, G5 is characterized by low fault density, moderate coal seam gas content, and competent roof lithology, which reduces the probability of uncontrolled gas migration. Additionally, its efficient gas drainage system, marked by early drainage initiation and consistent residual gas extraction, minimizes gas accumulation in the goaf. Well-designed ventilation pathways and effective goaf sealing further enhance gas dispersion, contributing to a stable subsurface environment and reducing the likelihood of secondary hazards such as spontaneous combustion or surface subsidence.

In contrast, G4 (Pingdingshan No. 8 Mine) and G6 (Tiefa Mine) recorded the lowest C_i values, indicating a significantly higher risk of gas enrichment. These zones are situated in structurally complex regions with dense fault networks and fragmented overburden, creating multiple gas migration pathways and limiting the effectiveness of sealing operations. G4 suffers from high coal seam gas content combined with a steep mining gradient and slow face advance, increasing emission rates that outpace the drainage system’s capacity. G6 displayed similar issues, with delayed gas drainage deployment and poor ventilation routing contributing to localized methane accumulation in closed-off sections of the goaf. Both areas also presented evidence of ecological vulnerability, including reduced vegetation cover and shallow groundwater depth, indicators of surface disturbance linked to underlying gas activity.

Mid-ranked zones such as G1 (Jincheng Western Mine,) and G3 (Huainan Panji Mine) exhibited moderate risk levels. These sites displayed intermediate fault development and acceptable ventilation performance but showed inconsistencies in drainage operation timing and residual gas management. While geologically stable, G2 (Pingshuo East Transition Zone) faced challenges with airflow distribution due to its open-pit-to-underground transition layout, limiting the effectiveness of conventional ventilation strategies. Overall, the results demonstrate that gas enrichment risk in goaf areas is not determined solely by a single factor, but rather by a complex interplay among geological structures, mining parameters, drainage efficiency, and ecological context. Zones with dense faulting, poor sealing, and delayed drainage consistently correlated with higher risk levels, while those employing proactive gas control and ventilation practices achieved lower risk scores. These findings underscore the necessity of tailoring mitigation strategies to each site’s specific geotechnical and operational conditions. High-risk zones such as G4 and G6 require intensified gas drainage protocols, real-time monitoring systems, and re-engineered airflow configurations. At the same time, low-risk areas like G5 serve as valuable benchmarks for best practices in integrated gas and land degradation management (Fig. 5).

Fig. 5

Comparison of normalized distances to ideal, non-ideal, and relative closeness values for gas enrichment risk ranking across six goaf areas.

Comparison with historical monitoring and events

To provide an external check on the model results, the ranking of the six goaf zones (G1–G6) was compared with available historical monitoring records and event information from the mines, including gas overlimit alarms, micro seismic or roof movement observations, and documented surface subsidence anomalies^46,47. Using historical monitoring and event data as independent evidence is a common practice in coal mine gas and stability studies, where numerical or risk models are often evaluated against measured gas concentrations, alarm statistics, and field observations before being used for decision support^48,49. Gas migration and emission models in goaf areas are frequently checked against measured gas composition and concentration profiles. At the same time, microseismic monitoring and subsidence measurements are widely used to assess roof caving, goaf stability, and mining-induced deformation⁵⁰. For the present case, the zones identified as highest risk by the AHP–TOPSIS framework (G4 and G6) correspond to areas where mine operators reported the most frequent gas over-limit alarms and pronounced subsidence or micro seismic responses associated with roof movement during the evaluation period. In contrast, the lowest-risk zone (G5) has no recorded gas exceedance events and only minor, localized subsidence, with no reported safety incidents in the same timeframe. The intermediate zones (G1–G3) exhibit occasional alarms or localized deformation, consistent with their intermediate composite risk scores. Although the available records do not constitute a complete statistical database and may be affected by reporting and instrumentation changes over time, the qualitative agreement between independent monitoring/event data and the model-based rankings is consistent with validation practices recommended in recent coal and gas risk assessment research. It provides additional support for the predictive validity and practical usefulness of the proposed framework.

Sensitivity analysis

A comprehensive sensitivity analysis was conducted to evaluate the robustness and reliability of the AHP-TOPSIS results. Given that the AHP methodology relies on expert-derived weightings, which may introduce subjectivity or variability, it is essential to assess the stability of the final rankings of goaf zones when input weights are varied. Sensitivity analysis provides insight into the model’s behavior under alternative weighting scenarios and helps identify which criteria most influence the final risk ranking. Five scenarios were developed for the analysis by systematically altering the weights of the main criteria while keeping the relative ratios of sub-criteria constant within each leading group. Each scenario was designed to reflect a plausible shift in expert opinion or focus, such as greater emphasis on geological features, operational factors, or ecological risk.

• Scenario 1 (Baseline): Uses original AHP-derived weights.

• Scenario 2 (High Geological Focus): The weight of Geological Conditions is increased by 20%, with proportional reductions in others.

• Scenario 3 (High Mining Focus): Increases Mining Parameters by 20%.

• Scenario 4 (High Ecological Sensitivity): Increases Ecological Vulnerability by 30%.

• Scenario 5 (Equal Weights): Assigns equal weights to all five main criteria.

The relative closeness values C_i for each goaf area (G1 to G6) were recalculated under each scenario using the modified weights in the TOPSIS model. Figure 6 summarizes the resulting ranks for all goaf zones across the five scenarios.

Fig. 6

Sensitivity analysis results showing the rankings of six goaf zones (G1–G6) under five weighting scenarios based on AHP-TOPSIS evaluation. The stability of extreme ranks (G5 as the lowest risk and G4 as the highest) demonstrates the model’s robustness, while moderate variations in mid-tier zones (e.g., G1, G2, G3, G6) highlight its sensitivity to changing criterion emphasis.

Figure 6 presents the results of a sensitivity analysis evaluating the robustness of the AHP-TOPSIS model rankings for gas enrichment risk in goaf areas. This analysis is crucial for understanding how changes in the relative importance (weights) of the five main criteria, Geological Conditions, Mining Parameters, Ventilation Characteristics, Gas Emission Rates, and Ecological Vulnerability, affect the prioritization of goaf zones (G1 to G6) based on their gas enrichment risk. The ranking of each goaf zone (from 1 = lowest risk to 6 = highest risk) under each scenario is then computed based on its relative closeness values (C_i) to the ideal solution. This provides a nuanced view of which zones consistently represent high or low risk, and which are sensitive to weight perturbations. Across all five scenarios, G5 maintains a consistent Rank 1, indicating it is the lowest-risk zone. This persistent top position underscores the robustness of its favorable conditions, which include low fault density, efficient drainage systems, and relatively stable ecological indicators. Its performance remains superior across all criteria, suggesting that G5 has a well-rounded safety profile. This makes G5 a strategic model for gas management and land rehabilitation under top-cutting and pressure-relief mining conditions.

At the opposite end of the spectrum, G4 consistently ranks as the riskiest zone (Rank 6) under all scenarios. This uniformity indicates structural and operational weaknesses that dominate its risk profile, including a dense fault network, delayed gas drainage, and poor ventilation configuration. The results affirm that no matter how the weighting is adjusted, G4 remains the most vulnerable, signalling an urgent need for remediation measures such as enhanced drainage, tighter sealing strategies, and ecological stabilization efforts. G6, while not always ranked last, occupies the second-highest risk position (Rank 2) in the first three scenarios. However, when ecological vulnerability becomes the focus in Scenario 4, G6 drops to Rank 5, reflecting its fragile surface conditions (e.g., reduced vegetation cover and shallow groundwater depth). This shift reveals the influence of surface ecological parameters on the overall assessment when they are prioritized. Likewise, G6 drops again in Scenario 5 (Equal Weights), maintaining Rank 5, reinforcing that its risk profile is especially pronounced when all factors are balanced. In contrast, G2 demonstrates moderate sensitivity to the weightings, slightly improving from Rank 3 in the first three scenarios to Rank 2 in Scenarios 4 and 5. This suggests that relatively better ecological and ventilation parameters counterbalance G2’s moderate geological and mining characteristics. Its improved rank when ecological factors are emphasized indicates that G2 may have less degraded surface conditions than higher-risk zones, making it a candidate for environmental conservation efforts rather than intensive engineering remediation. This behavior illustrates how the model can be tuned to different policy priorities: under a safety-dominated weighting, structurally complex zones are prioritized for engineering control, whereas under an ecology-focused weighting (Scenario 4), zones with fragile surface environments receive greater attention for conservation and preventive measures.

G1 and G3 also display stable rankings, typically in the mid- to lower-range. G1 maintains Rank 4 in the first three scenarios and improves to Rank 3 in the later scenarios. This gradual shift upward may be due to stronger ecological performance or more efficient ventilation practices becoming more influential as weightings change. G3, on the other hand, remains at Rank 5 in Scenarios 1 through 3 and moves to Rank 4 in Scenarios 4 and 5, again indicating modest sensitivity but consistently higher risk than G1 or G2. Overall, the sensitivity analysis reveals two significant insights. First, the model exhibits strong stability at the extremes: G5 and G4 remain the lowest and highest risk zones, respectively, under all weighting configurations. These zones can therefore be treated as robust management priorities, with G5 serving as a best-practice reference case and G4 as a critical area requiring sustained engineering and ecological remediation. Second, intermediate zones (G1, G2, G3, and G6) show varied sensitivity depending on which dimension of risk is emphasized. This variability is not only mathematically significant but also carries practical importance. For these “sensitive” zones, we recommend adopting conservative and differentiated strategies: (i) classifying them in the higher-risk category when rankings fluctuate between adjacent classes, (ii) implementing more intensive monitoring of gas concentrations, subsidence, and vegetation dynamics, and (iii) periodically re-evaluating their risk status under alternative weighting schemes that reflect evolving policy priorities (e.g., stronger emphasis on ecological vulnerability). In practice, the sensitivity results thus support a three-tier management approach: robust high-risk zones (e.g., G4, G6) for immediate engineering intervention, robust low-risk zones (e.g., G5) for maintaining current controls and using as benchmarks, and sensitivity-prone intermediate zones (G1–G3) for enhanced monitoring and precautionary measures. This level of diagnostic capability is essential for adaptive mine planning, effective ecological restoration, and the prevention of gas-related disasters in goaf areas.

A dual-axis scatter plot was developed using simulated values based on the results of this study to evaluate the correlation between gas enrichment risk and surface-level ecological degradation. The x-axis in Fig. 7 represents the gas enrichment risk scores (C_i) derived from the AHP-TOPSIS model for six goaf zones (G1–G6). The y-axes represent two ecological indicators: the Vegetation Cover Index (on the left) and Groundwater Depth in meters (on the right). These indicators were selected because they reflect ecological sensitive areas with low vegetation cover and shallow groundwater, which are typically more susceptible to surface degradation caused by subsurface instability. Vegetation cover values were derived from spatial vegetation coverage classes assigned to each goaf zone based on remote-sensing interpretation and field verification, and expressed as the proportion of vegetated area within each zone. Groundwater depth values were taken from hydrogeological monitoring and borehole records compiled for the case-study mines and aggregated to the same spatial units. The green circular markers indicate vegetation cover, while the blue squares denote groundwater depth. Both data series are accompanied by dashed trendlines representing linear regression fits. The results show a positive association between risk score and ecological indicators. As C_i increases (indicating a lower gas-enrichment risk), vegetation cover improves, and groundwater depth increases (suggesting a more stable, less degraded environment). This dual-axis analysis quantitatively supports the finding that goaf areas with higher gas risk tend to be accompanied by signs of ecological stress at the surface. In contrast, zones with better subsurface conditions (e.g., lower fault density, adequate drainage, and ventilation) show fewer ecological disturbances. These findings validate the inclusion of environmental vulnerability in the AHP-TOPSIS model and reinforce the link between subsurface gas dynamics and land degradation.

Fig. 7

A dual-axis scatter plot shows the relationship between gas enrichment risk scores (C_i) and ecological indicators across six goaf zones. Higher C_i values (indicating lower risk) correspond to increased vegetation cover and deeper groundwater levels. The trends confirm that gas enrichment in goaf areas is strongly associated with surface ecological degradation, validating the integration of subsurface and environmental criteria in risk assessment.

Discussions and limitations

While the combined use of AHP and TOPSIS is well established in the broader MCDA literature, the originality here lies in the problem-specific design of the criteria hierarchy and weighting scheme that jointly represent subsurface gas processes and surface ecological vulnerability under top-cutting and pressure-relief mining. The results underscore the effectiveness of this hybrid model in translating expert knowledge and field data into actionable insights for gas control and land degradation mitigation. However, while the model successfully identifies and ranks risk-prone areas, several critical discussion points and limitations must be addressed to contextualize its utility and guide future applications. The framework’s core strength lies in its ability to balance technical rigor with ecological sensitivity. The relative importance of criteria such as fault density, coal seam gas content, and emission rates aligns with industry understanding of gas behavior in goaf zones. At the same time, by incorporating ecological indicators like vegetation coverage and groundwater depth, the model moves beyond conventional mine safety paradigms to encompass broader environmental degradation dynamics. This dual focus is particularly relevant in regions where post-mining land use and ecological restoration are strategic priorities. As revealed in the preceding analysis, the implications for land degradation management are profound. High-risk zones such as G4 and G6, characterized by complex geological structures and operational shortcomings, showed clear signs of surface instability, including cracks, subsidence, and vegetation dieback. These zones exemplify how subsurface gas dynamics directly translate to surface degradation, reinforcing the need for integrated mitigation strategies. The model’s output enables targeted interventions such as:

• Accelerated deployment of directional drainage systems to minimize gas buildup.

• Re-engineering of ventilation circuits to ensure goaf flushing and prevent accumulation.

• Surface-level ecological measures, including revegetation and subsidence zone stabilization.

The AHP component of the model inherently relies on expert opinions, which, while valuable, are susceptible to bias, inconsistency, and contextual blind spots. Although consistency ratios (CR < 0.1) were used to validate inputs, the subjective nature of pairwise comparisons means the derived weights might still reflect underlying preferences or assumptions. In regions with different mining geologies or regulatory environments, alternative expert panels might produce diverging priorities, limiting the model’s direct transferability. The model was applied to six representative goaf zones from coal mines in China, selected for their diversity in geological and operational contexts. However, the availability and quality of input data, especially for ecological sub-criteria, varied across sites. This introduces uncertainty into the normalization and scoring processes. Additionally, while these sites are representative, they may not capture the full variability of mining conditions in other regions or mining systems (e.g., longwall vs. room-and-pillar). The model’s credibility is further supported by its qualitative agreement with independent mine records: zones ranked as high risk by the AHP–TOPSIS framework are the same areas where gas over-limit alarms and significant subsidence or micro seismic responses have historically been concentrated, whereas the lowest-risk zone shows an absence of such events over the evaluation period. However, the validation is constrained by the length and completeness of available records, and more systematic, long-term datasets would enable a more rigorous quantitative performance assessment in future work.

The AHP-derived weights are fixed once calculated and do not dynamically adjust based on real-time changes in mining operations, gas concentrations, or ecological feedback. In rapidly evolving goaf environments, this static weighting can reduce the model’s responsiveness: as sealing efficiency improves or degrades, ventilation layouts are adjusted, or residual gas emissions change over several years after mining, the relative importance of individual criteria for residual risk may also shift. In this study, the weights represent expert consensus for a given stage of top-cutting and pressure-relief mining, with a medium-term planning horizon. For long-term post-mining management, the framework should therefore be used in an iterative manner, in which (i) indicator scores are periodically updated using monitoring data (e.g., gas concentrations and flow rates, subsidence measurements, remote sensing based vegetation indices), and (ii) the weight vector is revisited at longer intervals or when management and policy priorities change. Such recalibration may be based on renewed expert judgment and, where data permit, supported by data-driven approaches (for example, entropy weighting or regression-based calibration linking criteria to observed risk outcomes). Without these updates, there is a risk that residual hazards in some goaf zones could be underestimated or that risk could be overstated in zones where remediation has been effective. Accordingly, the current weight system should be regarded as a baseline configuration that can be refined as new monitoring information and regulatory requirements become available. Moreover, the current AHP–TOPSIS framework does not explicitly account for the spatiotemporal evolution of gas concentration during specific events, such as sudden roof collapse or rapid fracture propagation. More complex dynamic approaches, for example, coupling numerical simulation of gas flow and rock deformation with MCDA, can provide higher-resolution insight into these short-term processes at the working-face scale. However, such models require detailed site-specific parameters and substantial computational effort, which limits their practicality for routine application across multiple mines. Under the current research objective, to compare and prioritize typical goaf conditions in several mines and support planning-level decisions, the static multi-criteria representation calibrated with expert knowledge and field data provides sufficiently reliable and interpretable risk zoning. Future work could therefore combine the present screening tool with targeted numerical simulations in the highest-risk zones to further refine local control strategies.

While incorporating ecological vulnerability into the model is a significant innovation, the selected indicators (e.g., vegetation cover, groundwater depth, land-use sensitivity) are proxies rather than direct measures of degradation caused by gas accumulation. This indirectness could understate or misrepresent the ecological consequences in certain zones, primarily where degradation is driven by factors beyond gas enrichment (e.g., heavy metal contamination, hydrological disruption). The classification thresholds for high, moderate, and low risk (based on Ci values) were empirically selected. Slight adjustments to these thresholds could reclassify some zones, affecting prioritization decisions. Future versions of the model could benefit from a more rigorous statistical or fuzzy logic-based thresholding approach to reduce ambiguity at category boundaries. Because site-scale quadrat surveys and detailed biodiversity or soil quality measurements were not consistently available across all mines, the present study relies on spatially consistent remote-sensing and monitoring datasets for ecological indicators; future applications could complement these with standardized quadrat-based field surveys to further refine the ecological component of the model. To avoid underestimating low-probability but high-consequence ecological events, we suggest a precautionary interpretation of the results: when a goaf zone exhibits high land-use sensitivity (C5.3) or ecological vulnerability, it should be treated as a higher management priority, even if its overall C_i value lies near the boundary between categories. Combined with the “High Ecological Sensitivity” weighting scenario in the sensitivity analysis, this provides decision-makers with an explicit mechanism to align risk classification and resource allocation with environmental policies and sustainable development goals.

Despite these limitations, the model provides substantial decision-making value, particularly when used as a screening tool or early warning system. It enables the stratification of mining areas based on integrated subsurface-surface risk profiles, guiding resource allocation, mitigation planning, and policy design. In particular, the sensitivity analysis results are used to distinguish between robustly classified zones, where intervention priorities are clear, and sensitivity-prone intermediate zones, where decision-makers are advised to adopt conservative classifications, strengthen monitoring networks, and periodically revisit the weighting scheme as management objectives or regulatory priorities evolve. Furthermore, the insights gained from the sensitivity analysis (Fig. 4) suggest that while the model is stable at the extremes (e.g., G5 and G4), it remains flexible and responsive in intermediate cases, an important quality for adaptive risk management.

In practice, this model supports a preventive rather than a reactive approach to mining safety and land rehabilitation. By identifying risk zones early and holistically, mine operators and regulators can implement context-specific strategies such as:

• Staggered drainage and ventilation planning during active mining phases.

• Real-time monitoring of methane levels and land deformation.

• Prioritization of ecological restoration efforts in zones showing both gas and degradation signatures.

Conclusions

This study developed and applied a robust multi-criteria decision-making (MCDM) framework combining the Analytic Hierarchy Process (AHP) with the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) to evaluate gas enrichment risk in goaf areas under top-cutting and pressure-relief mining conditions. By integrating geological, operational, ventilation, gas dynamic, and ecological criteria, the model enables a comprehensive assessment of subsurface hazard potential with direct relevance to land degradation control. The AHP method yielded a well-structured hierarchy of influence, with Geological Conditions (26.6%) and Mining Parameters (21.4%) emerging as the most critical factors in determining gas accumulation risks. Within these, Fault Density (11.5%), Coal Seam Gas Content (10.2%), and Mining Height (8.9%) were identified as dominant sub-criteria, reflecting their strong correlation with gas migration pathways, void formation, and emission potential. These insights underscore the need to prioritize structural geology and mining design when addressing goaf-related hazards. Applying the weighted criteria within the TOPSIS framework enabled the ranking of six representative goaf zones (G1–G6) across major Chinese mining regions. The results revealed clear spatial differentiation in gas risk:

• G5 (Tongchuan Yuhua Mine) was consistently ranked as the lowest-risk zone (Ci = 0.815), benefiting from favorable geological integrity, proactive drainage systems, and stable ecological surface conditions.

• In contrast, G4 (Pingdingshan No. 8 Mine) and G6 (Tiefa Mine) were identified as the highest-risk zones (Ci = 0.418 and 0.340, respectively), largely due to dense fault networks, delayed drainage, ineffective ventilation, and signs of surface ecological stress.

The ranking pattern is consistent with independent mine monitoring and event records, in which historically problematic goaf zones coincide with the model-identified high-risk categories. At the same time, the low-risk zone has remained free of gas exceedances and severe deformation during the evaluation period. These findings were further reinforced by a sensitivity analysis, which demonstrated that while rankings of high- and low-risk zones were stable under varying criteria weight scenarios, intermediate-risk areas showed moderate variability. This highlights the model’s robustness at the extremes and adaptability in transitional contexts, making it a powerful tool for targeted risk mitigation. Importantly, the model facilitated direct linkages between underground gas dynamics and surface land degradation indicators, offering an integrated basis for decision support in mine planning, safety management, and environmental restoration. The close alignment between zones with high gas enrichment and visible land degradation—subsidence, vegetation dieback, and ground fissures validates the ecological significance of subsurface risk. In operational terms, the AHP-TOPSIS framework contributes practical value in the following areas:

• Risk zoning and prioritization: enabling early identification of hazardous zones for targeted intervention.

• Decision-making transparency: combining expert knowledge with field data in a replicable structure.

• Multi-dimensional planning: accounting for engineering and environmental concerns in mine rehabilitation strategies.

Despite inherent limitations such as reliance on expert judgment and data availability constraints, the model demonstrates substantial potential for application in diverse mining contexts, particularly those facing dual challenges of gas control and post-mining land recovery. Future enhancements incorporating real-time sensor data, probabilistic modeling, and expanded ecological indices would further strengthen its predictive and management capacities. In addition, the framework is designed to be re-applied with updated indicator scores and, when necessary, revised AHP weights derived from new monitoring data and evolving policy or sustainability priorities, so that the initially static model becomes a periodically updated decision-support tool within an adaptive risk management cycle. At the same time, the AHP–TOPSIS framework is not intended to replace high-fidelity numerical simulations of gas migration. We do not claim methodological superiority over such dynamic models; instead, we show that, for the present objective of cross-mine comparison and prioritization of goaf zones under typical top-cutting and pressure-relief conditions, a static, data-efficient MCDA approach is sufficient to support reliable and transparent decision-making. In this sense, the proposed model complements detailed simulations: it provides regional-scale screening and factor diagnosis, while numerical models can be reserved for in-depth analysis of the most critical zones. Compared with prior MCDM applications in mining that address either discrete safety hazards or environmental impacts in isolation, the proposed framework provides a unified gas enrichment, land degradation risk assessment tool explicitly configured for goaf areas under top-cutting and pressure-relief mining. This shifts the role of MCDA from generic multi-factor aggregation to a mechanism-informed, site-adapted decision support system for sustainable mine and land management. In conclusion, the proposed hybrid AHP-TOPSIS model advances the methodological toolkit for evaluating gas enrichment risk and supports the broader goal of sustainable mining by integrating safety, operational efficiency, and ecological resilience into a unified risk management strategy. Its deployment can significantly enhance proactive hazard mitigation and land degradation prevention in goaf-prone mining environments.