Exploring the spatial characteristics of abandoned mining sites at the urban scale using a case study of Handan, China

Abstract

Research on spatial characteristics is the foundation for exploring the regeneration pathways of abandoned mining sites (AMSs) at the urban level. It aims to reveal their distribution status and regeneration laws, providing maximum support for the subsequent research stages such as the regeneration priority evaluation, direction decision-making, and strategy formulation. However, there are two important problems in current research on spatial characteristics: the lack of comprehensive and systematic research synthesis , which makes it difficult to grasp the study’s focus, significance, and methodology; and the lack of targeted research on AMSs. In most cases, the AMS is only a part of the research object and is commonly found in other fields, such as brownfields, resulting in conclusions that fail to match the AMS’s uniqueness accurately. To address these issues, this study advances research from both theoretical and practical perspectives. Theoretically, by synthesizing the research findings of AMSs and related fields (e.g., industrial heritage, mining heritage, brownfields, and urban vacant land) through the WOS and CNKI databases, we establish a theoretical system containing three dimensions of spatial structure, distribution and correlation characteristics, laying the foundation for comprehensive and systematic research. Practically, a case study on AMSs was conducted in Handan, China. This study revealed that, in addition to distance to urban areas, other urban location factors had little driving force on the regeneration priority of AMSs. This result differs greatly from those of brownfield and other fields, indicating that AMS regeneration is more likely driven by internal factors. This provides a basis for the selection of indicators for regeneration priority evaluation. In addition, we demonstrate a dynamic evaluation logic to evaluate the rationality of the regeneration progress of AMSs. This method is more scientific than the previous evaluation methods, which are based on regeneration quantity or area. On the basis of differences in regional evaluation results can guide the differentiated formulation of regeneration strategies. Overall, this study’s theoretical system and empirical findings will provide valuable references for research on spatial characteristics and subsequent research stages.

Introduction

Abandoned mining sites (AMSs) refer to lands once damaged by mining activities that are now abandoned or difficult to use without treatment. As urban diseases (congestion, pollution, functional chaos, etc.) are exposed and reflected globally, the demand for high-quality urban environments is continuously increasing. Given the limited land resources, relying on traditional land expansion to meet urban demands is becoming increasingly infeasible and unsustainable¹. In this context, urban renewal has emerged as a key strategy to alleviate urban issues, particularly in resource-based cities. Among various urban land types, AMSs have become critical urban renewal targets because of their large number, extensive area, severe environmental degradation, and negative impact on urban images. Their holistic research at the urban scale has also attracted increasing attention. In terms of other policies, research perspectives also show a trend of shifting from single-site to large-scale remediation stages. These include the Brownfields Area-Wide Planning Program issued by the U.S. Environmental Protection Agency (2012), the Infrastructure Strategy under the European Green Deal issued by the European Union (2020), and the 14th Five-Year Plan for Ecological Restoration of Mines in Beijing issued by China (2021). These policies overcome the limitations of site-scale research and provide a foundation for the holistic study of AMSs. However, current research has focused primarily on the site scale, with themes centered on heavy metal pollution management and landscape planning and design. From the construction of the world’s first influential mining park, Mont Chaumont Park in France in 1867, to the completion of the Sydney Olympic Games site in 2000, and up to the present day, over 150 years of research and practice have driven the maturity of site-scale research on AMSs. Relatively speaking, holistic studies of AMSs at the urban scale are noticeably lacking. Research in this area is not only limited in quantity but also scattered and lacking cohesion, thus failing to form a complete regeneration pathway for controlling and guiding the regeneration of AMSs. This presents a practical challenge for governments aiming to promote AMS regeneration from an urban perspective.  Therefore, if site regeneration is promoted blindly, it can lead to more problems, such as conflicts with existing plans, increased spatial inequality, high restoration costs, and insufficient alignment with public needs¹. To avoid these problems, it is crucial to construct a regeneration pathway to guide the scientific argumentation of AMS prioritization, direction, and strategic planning to achieve the rationalization and maximization of the benefits of AMS regeneration. In the process of path construction, advancing spatial characteristics research is the foundation and first task. It aims to reveal their spatial distribution status and regeneration laws, thereby providing essential theoretical support for subsequent stages such as the regeneration priority evaluation², direction decision-making³, and strategy formulation^4,5. If spatial analysis is missing or the research depth is insufficient, subsequent research stages may lack a support base and lead to suboptimal outcomes.

To clarify the research progress on the spatial characteristics of AMSs, this paper reviews and summarizes journal literature from two major databases—Web of Science (WOS) and China National Knowledge Infrastructure (CNKI)—with English used as the search language for WOS. The literature search included publications up to July 2024. The search terms encompassed five fields: industrial heritage, mining heritage, brownfields, abandoned mining sites, and vacant land. The search terms cover many fields because these concepts overlap, with AMSs commonly found within these fields of study. Taking industrial heritage as an example, the Tatachilla Charter clearly defines industrial heritage as including mining sites for ore extraction and smelting, as well as sites for energy production and transmission. Compared with general mining site ruins, industrial heritage includes the most representative, unique, and significant parts, namely mining heritage. Moreover, many industrial heritage studies do not meet the strict definition of heritage but instead rely on scholars’ subjective identification. Therefore, it is reasonable to include industrial heritage as one of the search terms. However, considering that industrial heritage, brownfields, and urban vacant land also contain other research objects, only the literature that explicitly contains AMSs will be screened out and analyzed to ensure the consistency of the search objectives. The search terms used were as follows: (spatial pattern* OR spatial feature* OR spatial characteristic* OR spatial distribution*) AND (Industrial heritage* OR mining heritage* OR brownfields* OR vacant land* OR abandoned mine* OR abandoned mining site* OR abandoned mining land* OR mining wasteland*). A total of 136 relevant studies were identified—50 from WOS and 86 from CNKI—with the majority published after 2010². This suggests that research on the spatial characteristics of AMSs remains in its early stages. On the basis of a general review of the literature, two key issues emerge. First, most studies focus on only one aspect of spatial characteristics and lack a comprehensive and systematic analysis. This fragmented research has made it difficult for us to identify the focus, significance, and appropriate methodologies for studying spatial characteristics, thereby impeding deeper research. For example, scholars such as Fu⁵, Filip⁶, and He⁷ have concentrated on existing abandoned mining sites (EAMSs), classifying their spatial patterns by exploring their positional relationships with built-up areas to formulate differentiated regeneration strategies. In contrast, Doick⁸, Ruelle⁹, and Chen¹⁰ focused on regenerated abandoned mining sites (RAMSs), examining changes in benefits after regeneration to evaluate the overall success of the regeneration projects. Some scholars, such as Frantál², Schneider¹¹, and Turečková¹², have focused on comparing the spatial distributions of RAMSs and EAMSs and their relationships with urban location factors to identify key regeneration drivers. This can help in the selection of indicators to evaluate regeneration priority and the direction of regeneration decision-making. In terms of research methodology, the literature shows wide variation: some scholars rely on qualitative descriptions¹³, whereas others employ quantitative techniques such as the nearest-neighbor index or Voronoi polygon analysis¹⁴. However, there are many research methods, and their similarities and differences have not been elucidated by anyone, leading to a growing trend of misuse among similar approaches¹⁵. Second, targeted studies on AMSs remain scarce^11,16,17. In most cases, the AMS is only a part of the research object and is commonly found in other fields, such as brownfields^{2,5,12,15,18,19}, which often cover diverse research objects, such as industrial, residential, and agricultural wastelands. Therefore, although integrated research across different research objects can reveal certain general laws, the findings often fail to match the uniqueness of individual site types, such as AMSs. This highlights the urgent need for targeted studies focused on AMSs to improve the accuracy of the results.

To address two key problems, this study is divided into two parts: (1) establish a comprehensive and systematic theoretical system by drawing on broader research from related fields such as industrial heritage, thereby guiding the in-depth analysis of spatial characteristics and overcoming the limitations of previous fragmented studies; and (2) conduct a targeted study on AMSs in Handan city, China, on the basis of` this theoretical system, to demonstrate the application of the theoretical system and explore AMS uniqueness. The ultimate goal of this paper is to provide maximum support for the subsequent research stages of the regeneration pathway, such as the regeneration priority evaluation, by guiding a comprehensive and systematic investigation of spatial characteristics to uncover spatial regeneration laws.

Theoretical system

To clarify the research focus, significance, and methodology concerning the spatial characteristics of AMSs, a theoretical system was established by reviewing studies directly related to AMSs and incorporating research from industrial heritage and other associated disciplines. A total of 242 documents were collected. The analysis revealed that spatial characteristics research can be categorized into three components: spatial structure, distribution, and correlation characteristics (Fig. 1). These components are distinct yet complementary, collectively forming the theoretical system. Among them, spatial structure and distribution characteristics aim to clarify the composition and distribution of AMSs to form complete basic knowledge. The spatial correlation characteristics constitute the core part of the study, which aims to explore the regeneration laws of AMSs through an analysis of their relationship with surrounding urban environments, thereby providing a theoretical basis for the subsequent research stages.

Fig. 1

Theoretical system of spatial characteristics research. This is the first task of this study. We summarized the research content, significance, and methods related to the spatial characteristics of AMSs through a literature review. Based on this, we developed a theoretical system to guide practical research, which is divided into three aspects: spatial structure, distribution, and correlation characteristics. The first two serve as the foundation, while the last is the focus of research on spatial characteristics.

(1) The study of spatial structure characteristics aims to summarize the acquired data on AMSs (e.g., quantity, type, area, regional differences, functional changes, boundary patterns^20,21,22) through qualitative analysis to develop a preliminary understanding. (2) The study of spatial distribution characteristics aims to further strengthen the overall understanding of AMSs through quantitative analysis. It mainly involves the aggregation pattern, the degree of aggregation, and the distribution range^23,24,25,26. For example, the standard deviation ellipse can determine the main distribution range of AMSs. On the basis of the analysis of multiyear data, the standard deviation ellipse can also be used to the changing trend in the distribution range²⁵. (3) The study of spatial correlation characteristics is divided into two parts. The first is a regeneration driver analysis. This was determined primarily by comparing the differences in the locations of RAMSs and EAMSs within the city. This study of driving factors is further subdivided into two aspects: drivers of regeneration priority (temporal sequence) and drivers of regeneration direction. This is the core of the past spatial characteristics research. Various analytical techniques have been applied, including ANOVA²², correlation analysis²⁷, regression analysis²⁸, the Kruskal-Wallis test²⁹, pairwise and discriminant analysis¹¹, the landscape association index³⁰, and grey correlation analysis³¹. However, as noted in the introduction, most findings are derived from broad fields such as brownfields and may not be fully applicable to AMSs. The second is post-regeneration evaluation analysis, which focuses only on RAMSs and mainly assesses the regeneration benefits and the rationality of regeneration progress. With respect to regeneration benefits, owing to the vast number of urban AMSs and limited access to certain indicators, constructing a comprehensive city-wide evaluation system remains difficult. Consequently, current studies often rely on single indicators such as tax revenue, employment generation, return on capital³², biodiversity improvements³³, traffic conditions³⁴, and landscape quality⁹. While such one-dimensional evaluations have limitations, they still provide valuable insights into specific aspects of regeneration outcomes. Overall, the value of further in-depth research in this area has substantially declined as the field nears its inherent limitations. . The rationality of regeneration progress refers to whether the regeneration speed aligns with predetermined planning goals. In practice, most cities do not have clear regeneration targets, and the regeneration progress of the AMS has often been described in terms of the restored quantity or area. However, due to the varying levels of development across cities, the financial support available also varies. As a result, these two indicators cannot be used for horizontal comparisons between cities. The lack of reasonable evaluation standards has been a major challenge in spatial characteristic research. Future research may employ dynamic evaluation methods, such as the coupling coordination model and grey correlation analysis³⁵, to establish relationships between regeneration status and urban development levels. The evaluation logic is more scientific and can guide the formulation of regionally differentiated regeneration strategies. This kind of research is more common in environmental science and ecology; however, it remains underexplored in the field of AMSs and related areas ³⁶, which needs attention. In addition, methods used in various aspects of the study are also summarized to avoid the misuse of similar methods.

Materials and methods

Research framework

The theoretical system can guide comprehensive and systematic analyses of spatial characteristics. However, owing to limitations in data availability, practical research often cannot achieve complete coverage—this study is no exception. Based on the available data in Handan city and drawing upon the theoretical system, we developed a practical research framework for the case of Handan, as illustrated in Fig. 2. Regarding spatial structure and distribution characteristics, four key indicators—the quantity and type, regional differences, aggregation patterns, and degree of aggregation—are selected to conduct a comprehensive comparative analysis of all AMSs, RAMSs, and EMSs, with the aim of deepening fundamental understanding. For spatial correlation characteristics, Spearman correlation analysis and geographical detectors are used to explore the quantitative-spatial relationships between AMSs and urban location factors. The aim is to uncover as many driving factors as possible that promote the prioritization of mining brownfield regeneration. This will provide more evidence for selecting indicators in the subsequent regeneration priority evaluation. In addition, we explored the coupled and coordinated relationship between the progress of AMS regeneration and the urban development level of each district (county) in Handan city, using the coupling coordination degree model to assess the rationality of the AMS regeneration progress (fast or slow). Given the dynamic nature of urban development and the lag in related data—with the latest data available up to 2023—studying the coupling coordination relationship for 2023 lacks practical significance. Therefore, this study first predicts the urban development level of Handan city in 2026 via the double exponential smoothing method on the basis of data from 2015 to 2023. The coupling coordination relationship is subsequently calculated to provide a scientific basis for formulating differentiated regeneration strategies for AMSs in Handan city.

Fig. 2

Research framework. We selected indicators based on the theoretical system and the available data to form a research framework using Handan city as a case study. In terms of spatial structure and distribution characteristics, four indicators—quantity and type, regional differences, aggregation patterns, and degree of aggregation—were chosen to analyze AMS spatial structure and distribution. For spatial correlation characteristics, Spearman correlation analysis and geographical detectors explored the quantitative-spatial relationships between AMSs and urban location factors. The coupling coordination degree model analyzed the relationship between the regeneration progress of AMSs and the urban development level, assessing the rationality of AMS regeneration progress in Handan’s districts.

Study area

Handan city is located in the Liaoxi–Taihang metallogenic belt in China, which is rich in coal and iron resources. In the 1970 s, China proposed the idea of “building the Handan—Xingtai industrial base with Handan as the center to create a Chinese version of the Ruhr District”. This promoted the comprehensive exploitation of Handan’s mineral resources. However, in the 21 st century, with the implementation of China’s safety and capacity reduction policies, most mining sites were forced to shut down in China, making Handan one of the cities with the highest concentration of AMSs. Consequently, the region faces significant potential issues, which are the primary reasons for selecting this area for research. As the eastern part of Handan lacks mineral resources, the study scope is limited to the downtown area (Fuxing District, Congtai District, Hanshan District), Fengfeng Mining Area, Yongnian District, Wu’an city, She County, and Ci County, with an area of 6,196.9 km² (Fig. 3). The map shows that mineral resources are closely associated with mountain ranges and are mostly distributed in a belt-like pattern. Along with mountains, water bodies, and urban areas, they shape the spatial pattern of the study area, providing a foundation for interpreting the results of the subsequent spatial characteristics.

Fig. 3

Study area. We clarified the study area’s scope, the distribution of mineral resources, and their spatial relationships with mountains, rivers, and the city, providing a foundation for interpreting the results of the subsequent spatial characteristics analysis.

Methods

Nearest neighbor index

This index reflects the spatial proximity of the AMSs. The balanced or uneven distribution of AMSs can be divided into three aggregation patterns: for R = 1, the AMSs are randomly distributed; for R > 1, the AMSs are uniformly distributed; and for R < 1, the AMSs are agglomerated. The formula and principle refer to the literature¹.

Kernel density Estimation

It is used to further corroborate the results of the calculation of the nearest-neighbor index. Compared with other methods, it has the advantages of being simple, fast, able to overcome administrative boundary constraints, and able to intuitively reflect the magnitude and specific location of point element clustering in space. The formula and principle can be found in the literature⁵.

Spearman correlation analysis

Correlation analysis is used to examine the relationship between AMSs and urban location factors and the strength of this relationship, providing a basis for subsequent geographical detector analysis to identify key regeneration drivers. The value of the correlation coefficient (CC) is within the [−1, 1] range, where higher absolute values denote more notable relationships. While correlation analysis is not as in-depth as regression analysis is, it can efficiently determine the strength of relationships from the difference in the CC values between RAMSs, EAMSs, and urban location factors. Given its suitability for this study, it was adopted. In geographic research, urban location factors are typically categorized into two major groups comprising nine subcategories—economic and social factors (population, transportation, resources, military, religion, and politics) and natural factors (topography, climate, and rivers). On the basis of this classification and the principle of data availability, appropriate indicators were selected to assess location factors (Table 1). These indicators have also been widely recognized in previous studies as key drivers of site regeneration, justifying the choice. Given the limited relevance of military and religion in Handan and the minimal intracity variation in climate, these three factors were excluded from the analysis. The X7 indicator was included because of its perceived strong influence on site regeneration.

Table 1 Evaluation indicators of urban location factors. Selected location factors. In geographic research, urban location factors are typically categorized into two major groups—economic and social factors (population, transportation, resources, military, religion, and politics) and natural factors (topography, climate, and rivers). Based on this classification and the principle of data availability, seven appropriate indicators were selected to assess location factors.

X1 was obtained from the Earth Resources Data Cloud (http://www.gis5g.com/data/rksj?id=1562), with a spatial resolution of 1 km in 2020. X2 (expressways and first- and second-level roads), X5, and X6 data were obtained from Bigemap. X3 was the sum of national, provincial and municipal cultural relic protection units and renowned towns and villages in Handan, derived from a government-published list. The calculation of X4 is centered on each district government office building, which is obtained through Google Maps. In this study, X8 was the ecological protection red line range, obtained from the Handan ecological protection red line distribution map.

Geographical detector

On the basis of the correlation analysis, the driving forces of seven location factors were verified via a geographical detector. The magnitude of the driving force is represented by the statistical value q, which ranges from [0, 1], with larger values indicating a stronger driving force. The formula and principle can be found in the literature²⁸. Unlike in previous studies, the dependent variable is set to only two values: 0 for EAMSs and 1 for RAMSs.

Double exponential smoothing

Double exponential smoothing is used to forecast the urban development level of Handan’s districts in 2026, providing essential data for subsequent coupling coordination analysis. This method smooths time series data by assigning different weights to reveal underlying trends. Compared with other time series forecasting methods, it is computationally efficient, requires minimal data, and accurately captures linear trends. The calculation process consists of two steps. First, the primary and secondary exponential smoothing values of the urban development level from 2015 to 2023 are computed. Then, on the basis of these smoothing values, the secondary exponential smoothing model is applied to predict the urban development level scores for 2026. The formulas for the first and second steps are shown below.

$$\:\begin{array}{c}{{S}}_{{t}}^{\left({1}\right)}{=}{{\alpha}}_{{1}}{\times}{{y}}_{{t}}{+}\left({1-}{{\alpha}}_{{1}}\right){\times}\:{{S}}_{{t}{-1}}^{\left({1}\right)}\end{array}$$

(1)

$$\:\begin{array}{c}{{S}}_{{t}}^{\left({2}\right)}{=}{{\alpha}}_{{2}}\:{\times}\:{{S}}_{{t}}^{\left({1}\right)}{+}\left({1-}{{\alpha}}_{{2}}\right)\:{\times}\:{{S}}_{{t}{-1}}^{\left({2}\right)}\end{array}$$

(2)

where S_t⁽¹⁾ represents the primary exponential smoothing value for year t (with 2015 as the reference year when t = 1); S_t⁽²⁾ represents the secondary exponential smoothing value at time t; α₁ and α₂ are smoothing coefficients ranging from [0,1], optimized through trial and error for better predictive performance, with values set at 0.5 and 0.6, respectively; y_t is the actual development level score for year t; S_t−1⁽¹⁾ is the primary exponential smoothing value for year t − 1, where the smoothed value for 2015 equals the actual development level score of that year; and S_t−1⁽²⁾ is the secondary exponential smoothing value for time t − 1.

$$\:\begin{array}{c}{{y}}_{t+T}{=}{{a}}_{{t}}{+}{{b}}_{{t}}{\times}{T}\end{array}$$

(3)

$$\:\begin{array}{c}{{a}}_{{t}}{=}{2}{\times}{{S}}_{{t}}^{\left({1}\right)}{-}{}{{S}}_{{t}}^{\left({2}\right)}\end{array}$$

(4)

$$\:\begin{array}{c}{{b}}_{{t}}{=}\frac{{\alpha}}{{1-}{\alpha}}{\times}{{[}{S}}_{{t}}^{\left({1}\right)}{-}{}{{S}}_{{t}}^{\left({2}\right)}]\end{array}$$

(5)

where y_t+T represents the secondary exponential smoothing forecast for the year T (after 2023), while the forecast values within the 2015–2023 time series are denoted as a_t. a_t is the intercept of the secondary exponential smoothing model, and b_t is the slope of the model. The model’s predictive accuracy is evaluated via the root mean square error (RMSE), where a smaller value indicates a better fit.

Coupling coordination degree model

The coupling coordination degree model is used to analyze the alignment between the 2023 regeneration progress of AMSs in various districts of Handan and their projected development levels in 2026 to formulate targeted regeneration strategies through 2026. Multiple models can be used to explore the relationships between two or more systems^35,42,43. The coupling coordination degree model is the first choice, as it can further illustrate the level of coordinated development between two systems⁴⁴. The formula and principles are as follows.

$$\:{{U}}_{{2}}\:{=}\:\sum\limits_{{j}{=1}}^{{m}}{{w}}_{{j}}{{x}}_{{ij}}{}$$

(6)

$$\:{C}\:{=}\:{2}\sqrt{\frac{{{U}}_{{1}}{{U}}_{{2}}}{{{(}{{U}}_{{1}}{+}{{U}}_{{2}}{)}}^{{2}}}}{}{}$$

(7)

$$\:{}{D}\:{=}\:\sqrt{{C}{\times}{T}}\text{}$$

(8)

$$\:{T}\:{=}\:{{\alpha}}_{\text{1}}{{U}}_{{1}}\text{+}\text{}{{\alpha}}_{\text{2}}{{U}}_{\text{2}}{{\alpha}}_{\text{1}}\text{+}\text{}{{\alpha}}_{\text{2}}\:=1$$

(9)

where U₁ represents each district’s score for AMS regeneration progress in 2023, and U₂ represents the projected urban development level of each district in 2026. m is the number of indicators, w_j is the weight of indicator j, x_ij is the normalized value of indicator j for district i, and C is the coupling degree, which indicates the magnitude of the correlation between the U₁ and U₂ systems. The closer the U₁ and U₂ values are, the larger the C value is, indicating a stronger correlation between the regeneration progress and the urban development level. The Formula (7) is the revised formula⁴⁴, which enhances the differentiation of C values while avoiding errors where two different data sets, such as 0.1 and 0.2 or 0.4 and 0.8, yield the same result. U_min is the smaller value between U₁ and U₂. D is the coupling coordination coefficient. The larger D is, the more closely aligned the progress of AMS regeneration is with the development level of each district and, thus, the higher the degree of coordinated development between the two. The D is divided into 10 levels⁴⁵. Based on the sizes of U₁ and U₂⁴⁶, each level can be further divided into 3 subcategories (Table 2). T represents the coordination index between C and D. α₁ and α₂ are the weights of U₁ and U₂, usually equally set at 0.5.

Table 2 Classification of coupling coordination levels. The coupling coordination level is divided into 10 levels based on D value, and each level is further subdivided into three sublevels based on the relationship between U₁ and U₂.

Evaluation system

The evaluation systems for AMS regeneration progress and urban development level were constructed sequentially to provide the necessary data for the study of the coupling coordination relationship. The former consists of two indicators: quantity and proportion of RAMSs (Table 3). The latter comprises three subsystems: economic, social, and ecological. Indicator selection was based on importance and availability, primarily referencing key indicators from the Handan Municipal Statistical Yearbook and other government reports (Table 4). The weights of the indicators in both tables are determined using the objective entropy method, as outlined in⁴⁶.

Table 3 Evaluation system for the AMS regeneration progress. It includes two indicators: two indicators: quantity and proportion of RAMSs.

Table 4 Evaluation system of urban development level. The system consists of three subsystems: economic, social, and ecological. Indicators were selected based on importance and availability, referencing the Handan municipal statistical yearbook and other government reports.

The data for B11–B24, D11, and D12 are from the Handan Municipal Statistical Yearbook; the data for C11–C22 are from the Hebei Statistical Yearbook; and the data for D21 and D22 are from the National Qinghai–Tibet Plateau Scientific Data Center (https://data.tpdc.ac.cn/home). This study compiles data from 2015 to 2023 for the above indicators, aiming to forecast the urban development level of each district in 2026.

Data

In addition to the data presented in Tables 1 and 4, this study also requires access to AMS data for the calculations in Table 3. The original data come from the National Mining Land Database 2021, which includes all mining enterprises in China that still existed in approximately 2004, with a total of 1,045 in Handan. Each mining enterprise contains basic attribute data such as spatial coordinates, names, mineral types, and utilization statuses (unused, active, and closed). However, there are issues such as duplicate entries, inaccurate utilization status (regenerated or not), and missing information, making data correction crucial. The study took 3 months to complete the data update in October 2023 by combining the literature, resulting in 771 AMSs. Therefore, U₁ is the regeneration progress score for 2023. Notably, the number of mining enterprises that are closed and not repaired is the number of AMSs in this study. Although some mining enterprises have multiple extraction points, they are commonly considered as a whole in actual planning since they are close to each other and share similar internal attributes. This study follows the same principle and does not separate them into more AMSs to avoid unnecessary work.

Results

Spatial structure characteristics

Quantity and type

The summary of the acquired AMS data reveals that Handan city has the greatest number of abandoned coal mines, totaling 388, followed by 249 abandoned iron mines and 134 abandoned quarries (including limestone, clay, and gypsum mines). Among these, 245 abandoned coal mines have regenerated, 88 abandoned iron mines have regenerated, and 62 abandoned quarries have regenerated.

Regional differences

In terms of the number of RAMSs, the Fengfeng Mining Area, Ci County, She County, and Wu’an city rank higher, with 93, 83, 82, and 70 regenerated sites, respectively. In contrast, Fuxing, Congtai, and Yongnian Districts have fewer RAMSs, with 41, 32, and 4, respectively (Fig. 4). With respect to the proportion of RAMSs, Congtai District has the highest evaluation, as all the AMSs have been fully regenerated, followed by Fengfeng Mining Area, She County, Fuxing District, Ci County, Yongnian District and Wu’an city, with proportions of 0.65, 0.64, 0.59, 0.54, 0.31 and 0.28, respectively. Fiscal support has played a significant role in this process. The municipal government has built a series of landscaping projects, such as the Zishan scenic area, in the western part of the downtown, which has rapidly restored the damaged ecological environment around the downtown. Other regions rely mainly on enterprises or natural slow recovery, with minimal government financial investment, resulting in significant differences in the regeneration proportions of AMSs across districts.

Fig. 4

Spatial structural characteristics of AMSs. Based on data statistics, we explained the types and quantities of all AMSs, RAMSs, and EAMSs in each district of Handan city, focusing on the proportion of RAMSs. This laid the foundation for the subsequent study on coupling coordination degree.

Spatial distribution characteristics

Aggregation pattern

The aggregation patterns of all the AMSs, RAMSs, and EAMSs in Handan were determined via the average nearest neighbor tool. The results show that they all exhibit an aggregated distribution in space. For example, the average actual nearest neighbor distance (R_i) for RAMSs is 701.6 m, which is smaller than the average theoretical nearest neighbor distance (R_e) of 1633.7 m. The nearest neighbor index R is 0.4, which is less than 1. The Z value is −21.44, which passes the 1% significance level test, and the significance level is P = 0. Therefore, it can be concluded that the RAMSs are spatially clustered and exhibit an aggregated distribution.

Aggregation degree

On the basis of the determined aggregation pattern, the aggregation degrees of all the AMSs, RAMSs, and EAMSs were visually analyzed via the kernel density tool. The results were classified into nine levels via the natural breaks method, and the kernel density maps were generated (Fig. 5).

The results indicated that all the AMSs exhibited a distinct spatial multicentered aggregation pattern (Fig. 5a): two high-density core areas formed in the western part of Congtai District and the central part of Ci County; two medium-density core areas formed in the southwest of Wuan City and She County; and two general-density areas were found in the northern part of Wuan City and the Fengfeng Mining Area. Comparing Fig. 5b with Fig. 5a, the high- and medium-density areas of RAMSs were generally consistent with the high- and medium-density areas of all the AMSs (except for the medium-density area in the western part of Wu’an city). A comparison of Fig. 5c with Fig. 5a reveals that the high-density core areas of the EAMSs have shifted compared with those in Fig. 5a. The high-density areas of the EAMSs are located more on the edge of the high-density core areas and within the medium-density core areas of Fig. 5a. These results reflect Handan’s accuracy in identifying key regeneration areas for AMSs in recent years,which is commendable.

Fig. 5

(a) The aggregation degree of all AMSs, (b) The aggregation degree of RAMSs, (c) The aggregation degree of EAMSs. We studied and compared the main distribution locations of all AMSs, RAMSs, and EAMSs, aiming to enhance the overall understanding of AMSs. These results reflected Handan city’s accuracy in identifying key regeneration areas for AMSs in recent years, which is commendable. 

Spatial correlation characteristics

Quantitative-spatial relationships with location factors

Through Spearman correlation analysis, the relationships between the AMSs and location factors were quantified (Fig. 6), and the results are presented in Table 5. First, we calculated the CCs between all the AMSs and the location factors to quantify the overall spatial correlation characteristics. We found that seven indicators were significantly correlated with the AMSs, with CCs greater than 0.7. These findings indicate that the original mining sites in Handan are closely linked to the socio-economic and natural environment. Next, by comparing the CCs of the RAMSs and EAMSs with the seven indicators, we found that the CCs between these two types of AMSs and X4 greatly differed (0.847 vs. 0.690), whereas the differences between the other indicators were not notable. This finding indicates that only X4 may play a driving role in the regeneration of AMSs.

Fig. 6

The spatial relationships between AMSs and location factors. Through Spearman correlation analysis, the relationships between the AMSs and location factors were quantified. The finding indicates that only X4 may play a driving role in the regeneration of AMSs.

Table 5 Quantification of the Spatial relationships between amss.and location factors. Through spearman correlation analysis, the relationships between the amss.and location factors were quantified. The finding indicates that only X4 May play a driving role in the regeneration of amss.

On this basis, a geographical detector was used to verify the driving forces of seven indicators. The results revealed that, except for X4, the q values of the other variables were all less than 0.1 (Table 6), confirming that the other variables had minimal driving effects on site regeneration. The above analysis indicates that external location factors have a very limited influence on the regeneration of AMSs, whereas internal site attributes may play a more critical role. This finding significantly differs from those of previous studies (as shown in Table 1, where these factors were identified as key drivers of regeneration priority in past research). This supports the conjecture made in the introduction that results derived from broader areas of research, such as brownfields, are not applicable to AMSs.

Table 6 Driving force (q value) of Spatial variables. We used a geographical detector to verify the driving forces of seven factors. The results revealed that, except for X4, the q-values of the other variables were all less than 0.1, confirming that the other variables had minimal driving effects on AMS regeneration.

Coupling coordination relationship with the urban development level

(1) Forecast the urban development level in 2026. First, based on Eq. (6) and Table 4, the actual development levels of each district from 2015 to 2023 were calculated. Then, via Eqs. (1)–(5), the predicted urban development levels for the same period were computed. A comparison between the predicted and actual values (Fig. 7) reveals a high degree of consistency in the trend. To evaluate the prediction accuracy, the RMSE between the predicted and actual values for seven districts was calculated, revealing a maximum RMSE of only 0.03. This indicates minimal differences between the predicted and actual values, demonstrating that the model is reliable for forecasting the urban development level of each district in 2026. The final forecast results are presented in Table 7.

Fig. 7

Comparison of predicted and actual urban development levels of Handan. (a) Fuxing District; (b) Congtai District; (c) Fengfeng Mining Area; (d) Yongnian District; (e) Wu’an City; (f) Ci County; (g) She County. We used double exponential smoothing to forecast the development levels of various districts in Handan from 2015 to 2023. By comparing the predicted values with the actual values, we found a high accuracy, which can be used to predict the urban development level in 2026, as U₂ values for subsequent research on coupling coordination degree.

(2) Determine the coupling coordination type. U₁ is calculated on the basis of the indicator weights (Table 3), and the quantity and proportion of RAMSs for each district obtained in Sect. “Regional differences” Then, the predicted urban development level scores for each district in 2026 (U₂) are combined, and the values of C, T, and D are sequentially calculated via Eqs. (7)–(9). The results are presented in Table 6.

Table 7 Results of coupling coordination types. In handan, the coupling coordination types are classified into three categories. The highest evaluation for their AMS regeneration efforts among the seven districts is for Fuxing district, Congtai district, Fengfeng mining area, and ci county. Yongnian district and wu’an City have received the lowest evaluations. Their coupling coordination type (barely coupling coordination–U₁ Lagged) reveals that wu’an and Yongnian have the financial resources and potential to promote more AMS regeneration, but the reality falls short.

(3) Analyze the coupling coordination types and formulate regeneration planning strategies. The coupling coordination types can be used to assess the rationality of AMS regeneration efforts in each district. In Handan, the coupling coordination types are classified into three categories from highest to lowest.Among them, Fuxing District, Congtai District, Fengfeng Mining Area, and Ci County have received the highest evaluation for their AMS regeneration efforts. Their coupling coordination types are classified as slight coupling coordination–U₁ advanced type and primary coupling coordination–U₁ advanced type, indicating a favorable state of coordinated development. The U₁ advanced type suggests that these districts have prioritized AMS regeneration in urban development, allocating significant resources and achieving remarkable results. This aligns closely with China’s principle of “ecological priority and green development” and is commendable. In 2026 and for a long period, their focus can shift to other urban development areas. From a workload perspective, the regeneration efforts of AMSs in She County are just right. Its coupling coordination type is the U₁ and U₂ synergistic type, indicating that the AMS regeneration work and urban development level are in an ideal state of balanced development. In the short term, its focus can also be directed toward other urban development areas. However, given the continuous upward trend in the urban development level in recent years, She County still needs to continue advancing AMS regeneration to ensure long-term coordination with urban growth. Yongnian District and Wu’an city have received the lowest evaluations, especially Yongnian District. Their coupling coordination type (U₁ lagged) reveals that Wu’an and Yongnian have the financial resources and potential to promote more AMS regeneration, but the reality falls short. This indicates that the efforts and the level of attention given by Wu’an city and Yongnian District need to be further enhanced. As shown in Fig. 5c, the high-density core area of AMSs in the western part of Wu’an city is likely to become a key area for future remediation efforts.

Discussion

Comprehensiveness and systematization of the theoretical system

Most current studies focus on only one aspect of spatial characteristics, with studies centered on the driving factors being the most prevalent². In contrast, comprehensive and systematic investigations into spatial characterization are relatively scarce. This defect makes it difficult for us to identify the research focus, significance, and appropriate methodologies, thereby impeding deeper research. While specialized studies on individual issues hold considerable academic value, comprehensive and systematic research on spatial characteristics is even more critical. As the foundation of urban-scale research, spatial characteristics are closely linked to the regeneration priority evaluation, direction decision-making, and strategy formulation, playing a critical supporting role (Fig. 1). However, the purpose of spatial characterization research is often overlooked, leading to fragmented studies and a lack of cohesion. Therefore, the first part of this study aims to construct a theoretical system for spatial characterization to clarify the research focus, highlight the significance of key elements, and guide methodological choices, laying the foundation for comprehensive and systematic research. To this end, we conducted an in-depth literature review on AMSs and relevant fields on the basis of the WOS and CNKI databases. Our analysis reveals that dividing spatial characteristics into three dimensions—spatial structure, spatial distribution, and spatial association—effectively encompasses the key elements of regeneration priority evaluation and other associated research stages. Specifically, spatial structure characteristics aim to summarize and organize the collected data on AMSs, forming a basic understanding. The spatial distribution characteristics aim to explore the distribution patterns of AMSs through quantitative analysis to deepen the overall understanding. As the core of this study, spatial correlation characteristics primarily investigate the drivers of regeneration, the benefits of regeneration, and the rationality of regeneration progress to support subsequent research. Although many existing studies emphasize the importance of spatial correlations among AMSs^28,47, they were excluded from the research system because of their limited relevance to downstream research stages. Additionally, research methods were systematically reviewed to prevent the misuse of similar approaches. Overall, the first part of the study effectively makes up for the theoretical system and provides a reference for the subsequent in-depth study of spatial characteristics.

Uniqueness of the Spatial characteristics of AMSs

The AMS is an important area of multidisciplinary intersection research. Over the years, scholars have conducted extensive investigations into mine restoration, generating a substantial body of academic work that provides a strong foundation for further study. However, most existing research focuses on the site scale, with relatively few holistic studies at the urban level, and spatial characterization also remains insufficiently addressed. Because we observed that spatial characterization studies targeting AMSs are rare^11,16,17, and more often, the AMS is only a part of the research object, which is commonly found in other fields, such as brownfields, resulting in conclusions that fail to match the uniqueness of AMSs accurately. To address this gap, the second part of this study takes a targeted research of AMSs in Handan city, aiming to reveal the unique spatial characteristics of the AMSs. The study revealed that this uniqueness is primarily reflected in the regeneration priority drivers. Apart from the factor ‘distance to urban areas’, other urban location factors had little driving force on the regeneration priority of AMSs. (Table 6). This indicates that internal site-specific factors are more critical in the regeneration process. Compared with findings from previous studies focusing on broader categories such as brownfields (Table 1), the results presented here differ significantly. This divergence is primarily due to the spatial consistency of the mineral resource distribution (Fig. 3). In Handan, mineral resources are concentrated mainly in the shallow mountainous zones, which are relatively distant from urban centers but closely connected to mountain systems. In the 1970 s, driven by national economic development policies, China entered a phase of large-scale mineral exploitation. This enhanced transportation infrastructure and population concentration tightly linked mining sites with the surrounding economic and social environments (see the second column of data in Table 5). Consequently, RAMSs and EAMSs are strongly associated associations with economic, social, and natural conditions, leading to the diminished influence of most urban location factors other than ‘distance to urban areas’. Compared with earlier studies, the second part of this research provides more accurate insights and further validates the necessity of conducting targeted studies on AMSs.

Limitations

This study has two main limitations. First, the results concerning the regeneration driving factors of AMSs require further validation. This study uses only Handan city as an example, which is insufficient. Future research should include analyses of more cities to validate the conclusions. Second, owing limited data accessibility, some studies outlined in the theoretical system could not be implemented, such as investigations of drivers influencing the regeneration direction. Some studies remain incomplete. For example, the evaluation system of the urban development level presented in Table 4, key indicators such as the forest coverage rate and green space coverage in built-up areas were excluded due to the difficulty of obtaining these data, especially continuous multiyear data, at the district and county levels in China. Data constraints also affected the definition of the U₁ value in this study. U₁ is composed of the quantity and proportion of RAMSs. Coupling this metric with the urban development level allows for a horizontal comparison of regeneration progress across districts, providing meaningful insights. However, since AMSs vary significantly in size, using only the quantity and proportion of RAMSs as the indicators may not fully and accurately assess the effectiveness of districts’ AMS regeneration efforts. A more scientific approach would include the area and area share of RAMSs in Table 3. Unfortunately, owing to a lack of statistical data, this study could not adopt this approach. Moreover, regarding regeneration priority drivers, regeneration is typically influenced by both external location factors and internal self-factors. This study focused only on the former, as internal factors—such as geohazard risks and residual contamination—were difficult to assess because the RAMSs have already been restored, making their original data hard to obtain. . We believe that the primary obstacle to advancing spatial characterization research lies in data acquisition. Although the theoretical system clearly defines the focus of the study, practical research can only be furthered if a comprehensive database is established with governmental support.

Conclusion

In contrast to previous studies that focused on the site scale, this paper adopts an urban-scale perspective to explore the overall spatial characteristics of AMSs, providing new perspectives and directions for related research. Research on spatial characteristics is the foundation for exploring the regeneration pathways of abandoned mining sites (AMSs) at the urban level. It aims to reveal their distribution status and regeneration laws, providing maximum support for subsequent research stages, such as regeneration priority evaluation, direction decision-making, and strategy formulation. However, two major problems persist in current research: the lack of comprehensive and systematic research, which makes it difficult to grasp the study’s focus, significance, and methodology; and the lack of targeted research on AMSs. In most cases, the AMS is only a part of the research object, which is commonly found in other fields such as brownfields, resulting in conclusions that fail to match the AMS’s uniqueness accurately. Overall, these issues weaken support for subsequent regeneration stages and reduce the accuracy of related outcomes. To address this, the present study makes improvements at the theoretical and practical levels. At the theoretical level, by synthesizing the research findings of AMSs and related fields (e.g., industrial heritage, mining heritage, brownfields, and urban vacant land) through the WOS and CNKI databases, we establish a theoretical system containing three dimensions of spatial structure, distribution, and correlation characteristics. This system clarifies the core focuses, theoretical significance, and methodological approaches of spatial characterization of AMSs and lays the foundation for comprehensive and systematic research. At the practical level, guided by the theoretical system and considering data accessibility, we conducted a case study in Handan city. In the study of spatial structure and distribution characteristics. we used qualitative analysis, nearest neighbor index, and kernel density estimation to compare and discuss the AMSs, RAMSs, and EAMSs from multiple aspects, including quantity, type, and kernel density concentrations to deepen the fundamental understanding. . In the study of spatial correlation characteristics, we conducted an in-depth exploration focusing on two key issues. First,  we examined the regeneration priority drivers. Through Spearman correlation and geographic detector analyses, we found that the key factors driving the regeneration of mining derelicts are likely to stem from internal factors. These findings suggest that key regeneration drivers may originate from the internal self-factors of AMSs. Thus, the regeneration priority evaluation system should emphasize the selection of internal indicators to more accurately identify high-priority regeneration sites. Second,  we evaluated the rationality of regeneration progress across districts in Handan. An analysis of the coupling and coordination relationships between the AMS regeneration progress and the urban development level, using the double exponential smoothing method and the coupling coordination degree model, revealed that four out of the seven districts performed relatively satisfactorily, whereas Wuan city and Yongnian District ranked lowest. Although Wuan city and Yongnian District possess the capacity to advance regeneration, the current outcomes are suboptimal, indicating a need for greater effort and focus. Combined with kernel density analysis of the EAMSs (Fig. 5c), the study concludes that the western and eastern parts of Wuan should be prioritized as key areas for future restoration. Overall, this study investigates the spatial characteristics of AMSs at both the theoretical and empirical levels. The proposed theoretical system and derived conclusions provide valuable references for promoting the regeneration of AMSs at the urban scale. The theoretical system proposed in this paper will lay the foundation for deeper research into spatial characteristics, and the numerous conclusions derived from multiple methods will provide the greatest support for subsequent steps of the regeneration pathway, such as the regeneration priority evaluation.