Green mining policies reduce carbon emissions in Chinese coal production bases

Abstract

Countries worldwide are pursuing efficient and clean energy extraction. In China, coal bases have implemented green mining policies for over a decade, but their environmental impact remains poorly understood. Here we compile raw coal production data and green mining policy records from 14 coal bases between 2004 and 2021 to estimate carbon emissions and quantify policy intensity. We assess the effect of these policies on emissions and find that, following national green mining initiatives, policy intensity increased more than tenfold, coinciding with a substantial reduction or slowdown in carbon emissions from raw coal production. Implementation of these policies is estimated to have reduced total emissions by 43.59 million tons annually. Differences in policy design, enforcement capacity, and technological foundations contribute to variability in emission reductions. These results demonstrate the critical role of policy interventions in guiding coal-dependent regions toward effective climate governance and provide insights for designing more scientifically grounded strategies to achieve cleaner and more efficient energy production.

Introduction

Global warming is an indisputable climate challenge that drives a range of energy and environmental issues¹. The international community recognizes anthropogenic climate change as a critical global concern, with energy extraction and consumption constituting substantial contributing factors². In 2020, the global production of mineral fuels, metal ores, and industrial minerals reached 17.3 billion tons, a 52% increase from 11.3 billion tons in 2000³. In 2021, coal-related CO₂ emissions reached a record value of 15.3 Gt⁴. The proposal at COP27 in 2022 to phase out all fossil fuels was rejected⁵, underscoring that reducing reliance on coal and other fossil fuels remains a protracted and uncertain process⁶. Consequently, coal mining (Fig. 1a) and consumption are expected to persist for the foreseeable future. Moreover, the global mining sector continues to face severe pollution, elevated safety risks, and low extraction efficiency. As global awareness of sustainable development and environmental protection increases, urgent policy actions⁷ are needed to reduce energy-related carbon emissions and advance climate goals. According to the Climate Change Laws of the World database, as of November 2025, there are 5008 climate change laws and policies globally (https://climate-laws.org/). International organizations, including the United Nations Environment Programme and the International Council on Mining and Metals, have advocated for the adoption of environmentally friendly mining practices to achieve breakthroughs in climate and environmental resilience, social performance, governance, transparency, and sustainable development innovation (https://www.icmm.com/en-gb/our-story/our-approach). Several countries are actively developing policies to promote carbon reduction, such as the UK Climate Change Act and Germany’s Renewable Energy Law⁸. In China, green mining policies have been established to promote energy conservation, emission reduction, and high-quality development of the coal industry⁹.

Fig. 1: Map of global coal mines and green mines (coal mines) in China.

a Global map of coal mining sites. b Map of green mines (coal mines) in China; the cut-out shows the South China Sea Islands.

China is a major coal producer and consumer, accounting for ~51.8% of global coal production and ~31.9% of global energy-related carbon emissions in 2023¹⁰. In that year, China’s raw coal production reached 4.71 billion tons, a 3.4% increase from the previous year¹¹. Coal remains a fundamental energy source in China¹², with its share of primary energy consumption expected to remain above 50% by 2025¹³. However, China’s coal mining industry continues to face challenges, including high energy consumption, elevated emissions, and severe pollution, leading to substantial environmental disturbances and substantial greenhouse gas emissions¹⁴. Therefore, the implementation of green mining policies is highly important for advancing mining technology, improving the mining environment, reducing pollution, and further promoting the realization of dual carbon goals and high-quality development. Initially established to ensure stable energy supply, energy bases now also undertake tasks such as supply security, intelligent mining, clean utilization, and ecological restoration. The coal base is a critical coal production area in China and is characterized by large coal enterprises that facilitate industrial agglomeration and enable coal transfer, power supply, coal chemical production, and comprehensive resource utilization. China has planned and established 14 large coal bases¹⁵, which together cover 32.78% of the country’s land area. Each coal base spans multiple cities and encompasses several national-level green mines.

China’s green mining policy evolved from the concept of “green mine”, which was initially proposed by the Ministry of Land and Resources in 2000¹⁶. In 2007, the Ninth China International Mining Conference explicitly advocated for the development of green mine⁹. By 2010, the Ministry of Land and Resources issued a policy document outlining the overarching framework, fundamental requirements, guiding principles, and key tasks for green mine construction¹⁷. Green mining has since become a national strategy (Fig. 1b). Currently, the specific objectives of green mining include achieving an ecologically sustainable mining environment, employing scientific mining methods, optimizing resource utilization efficiency, implementing digital management systems, and fostering harmonious mining community relations.

Current research on carbon emissions encompasses both macro- and microscales. At the macro scale, studies have focused predominantly on the regional¹⁸, national¹⁹, provincial, and municipal levels. From a research perspective, studies have investigated carbon emissions from energy consumption, enterprises²⁰, industries, or sectors (e.g., cement²¹, steel²², wastewater treatment²³, and housing¹⁷), as well as lifecycle emissions from mining activities (e.g., copper mining²⁴). Common carbon emission calculation methods include the top-down emission factor approach²⁵, remote sensing satellite monitoring^18,26, and bottom-up methods such as lifecycle assessment²⁷ and direct measurement. Additionally, future carbon emissions are often predicted using diverse models^28,29 and scenario simulations. Studies indicate that carbon emissions have remained high and have experienced substantial growth over the past decades. However, comprehensive and region-specific analyses of carbon emissions from raw coal production within particular types of mines in large-scale coal bases are lacking. Consequently, it is essential to systematically analyze the carbon emissions associated with the raw coal production process within these coal mine bases, which represent substantial clusters of coal mining activity in China.

Current research on the factors influencing carbon emissions can be broadly categorized into two categories. First, from the perspective of external interventions, studies analyze how policy-making affects carbon emissions. This includes examining the impact of national strategies on carbon emissions in cities or regions. Examples include the effects of climate policies on carbon reduction³⁰, the influence of fertility and retirement policies on emissions³¹, the role of mayors in urban carbon emission interactions³², and the effects of green fiscal policies³³ and electric heating policies on carbon emissions³⁴. Additionally, research has explored the interactions between policies³⁵ and the effects of optimizing policy levels on carbon emissions³⁶ to optimize policies. Second, from the perspective of endogenous (deterministic) factors, research has examined the role of variables such as the digital economy, technology³⁷, energy structure and intensity³⁸, building characteristics³⁹, consumption behaviors⁴⁰, and population and economic growth⁴¹ in driving carbon emissions, particularly in the context of energy consumption. Research methods for external interventions include primarily econometric models^32,33,41 and scenario simulation approaches^37,42. To analyze carbon emission impact factors from a production perspective, methods such as spatial statistical and spatial econometric techniques are typically used.

Many existing studies have examined the effects of public policies on energy-related carbon emissions at the urban or regional scale, revealing that policy interventions may generate heterogeneous and context-dependent outcomes. However, this body of work has focused primarily on energy consumption on the demand side, such as in cities, industries or households, while far less attention has been given to the production-side emissions of resource extraction sectors, particularly coal mining. In addition, previous research has rarely incorporated policy implementation intensity into empirical analysis. Policies are often treated as binary interventions rather than dynamic and evolving processes. This limitation obscures how variations in policy strength and enforcement across regions and over time influence emission outcomes. To address these research gaps, this study develops a comprehensive analytical framework that links coal production-related carbon emissions with the intensity of green mining policies across fourteen major coal bases in China from 2004 to 2021. To clearly analyze, we divided the 14 coal bases into four regional groups on the basis of China’s economic zoning: the northeast region includes the Mengdong base (MD), and the eastern region includes the Luxi (LX) and Jizhong (JiZ) bases, and the central region includes the Jinbei (JB), Jinzhong (JinZ), Jindong (JD), Lianghuai (LH) and Henan (HN) bases, and the western region includes the Shendong (SD), Shanbei (SB), Huanglong (HL), Yungui (YG), Ningdong (ND) and Xinjiang (XJ) bases. This grouping allows us to present the results more clearly and aligns more closely with China’s overall strategic layout for coal development, which aims to compress production in the eastern region, restrict expansion in the central and northeastern regions, and optimize development in the western region. Specifically, we (1) construct a text-based policy intensity index to quantify the evolution and heterogeneity of green mining policies, (2) estimate coal production-related carbon emissions before and after the introduction of national green mining policies and (3) apply interrupted time series analysis (ITSA) to identify the causal effects of policy introduction and policy strength on emission trajectories. This approach enables us to move beyond descriptive policy assessments by capturing the temporal dynamics of policy impacts, regional differences among coal bases and the direct emission outcomes of coal production. In doing so, our study provides the first quantitative evaluation of how green mining policies have affected the carbon emission patterns of China’s coal production area, an important but underexplored dimension of global low-carbon transition strategies.

Results

Carbon emissions from raw coal production in China’s coal bases

Overall, carbon emissions from coal production increased in China’s 14 national coal bases between 2004 and 2021, accompanied by the expansion of high-emission zones (Fig. 2a and Supplementary Fig. 1b). This trend is likely driven by the increase in raw coal production resulting from China’s energy resource endowment and energy structure.

Fig. 2: Spatiotemporal map of carbon emissions from coal bases in China.

a Spatial distribution map of carbon emissions from raw coal production at 14 coal bases in 2021; the cut-out shows the South China Sea Islands. b–o Statistical charts of carbon emissions from raw coal production in 14 coal bases from 2004 to 2021 (unit: 10,000 tons).

From a regional perspective, the MD in Northeast China experienced an overall increase in carbon emissions during the study period. Specifically, carbon emissions increased steadily from 2004 to 2012, reaching a peak of 70.4401 million tons in 2012. Carbon emissions then declined from 2013 to 2017, followed by a slight decrease and subsequent increase after 2018, with all changes remaining relatively small (Fig. 2b). In the eastern region, both JiZ and LX showed overall fluctuating downwards trends in carbon emissions over the study period, characterized by an initial rise followed by a gradual decline, although the fluctuations were more pronounced in the JiZ. With respect to the JiZ, carbon emissions remained relatively stable from 2004 to 2008, increased substantially from 2008 to 2011, reaching a peak of 29.4287 million tons in 2011, and then gradually declined thereafter (Fig. 2c). The LX maintained stable carbon emissions from 2004 to 2008, experienced a slight increase between 2008 and 2010—peaking at 16.5472 million tons in 2010—and then slowly and steadily decreased, reaching its lowest level by 2020 (Fig. 2d).

In the central region, carbon emissions increased overall across the five coal bases. The JB, JinZ, JD, and LH all showed fluctuating upwards trends, whereas the HN experienced a decline. The JB rose gradually from 2004 to 2012, reaching a peak of 69.7437 million tons in 2012 (Fig. 2e). Carbon emissions then declined from 2013 to 2016, followed by a slow increase from 2017 to 2020, and a slight decrease in 2021. The JinZ showed the greatest increase with pronounced fluctuations. In addition to a substantial decline in 2009, carbon emissions increased from 2004 to 2015, decreased again from 2015 to 2016, and then rose from 2016 to 2020, reaching a peak of 70.7470 million tons in 2020, followed by a slight decrease in 2021 (Fig. 2f). The changes in the JD were relatively stable. Apart from a notable increase between 2007 and 2008, annual increases or decreases were relatively small, and carbon emissions reached their maximum of 61.3869 million tons in 2021 (Fig. 2g). The LH experienced the smallest increase. The carbon emissions first increased, but then fluctuated downwards. From 2004 to 2012, the carbon emissions increased slowly and reached a peak of 29.7146 million tons. After 2012, the carbon emissions declined with fluctuations, with slight increases in 2015 and 2020 (Fig. 2h). The decrease in the HN was the greatest. The carbon emissions initially fluctuated and then gradually decreased. Except for a slight decrease in 2006, the carbon emissions increased from 2004 to 2009 and reached a peak of 64.5092 million tons in 2009. The carbon emissions subsequently continued to decrease, reaching their lowest level in 2021 (Fig. 2i).

In the western region, carbon emissions increased substantially overall. The SD increased greatly from 2004 to 2021. The carbon emissions first increased rapidly but then continued to increase with fluctuations, with two periods of decline. The carbon emissions rose continuously from 2004 to 2012, began to fluctuate in 2013, and reached a peak of 131.5074 million tons in 2019. The carbon emissions decreased in 2020 but increased again in 2021 (Fig. 2j). The decline after 2012 may reflect the influence of the green mining policy. Carbon emissions in YG increased with fluctuations. The carbon emissions fluctuated upward before 2013 and declined with fluctuations thereafter. The carbon emissions peaked in 2013 at 107.0090 million tons, decreased steadily from 2013 to 2018, and then increased slowly, remaining below the 2013 peak (Fig. 2k). The SB exhibited the greatest overall increase. Except for a relatively stable period from 2008 to 2016, the carbon emissions increased rapidly in all other years and reached a maximum of 56.5930 million tons in 2021 (Fig. 2l). The HL increase in HL was smaller. The carbon emissions increased from 2004 to 2012 and reached a peak of 29.7779 million tons, decreased from 2012 to 2013, and then remained generally stable (Fig. 2m). In XJ, the carbon emissions increased rapidly. The carbon emissions increased slowly from 2004 to 2012, remained relatively stable with slight fluctuations from 2012 to 2017, increased and then decreased between 2017 and 2019, and increased again after 2019, reaching a maximum of 21.4521 million tons (Fig. 2n). Among the six coal bases, the ND exhibited the most stable carbon emission pattern, with carbon emissions slowly fluctuating upwards from 2004 to 2015 to a maximum of 3.67 million tons, followed by a gradual decline thereafter (Fig. 2o). In 2004, the overall carbon emissions from raw coal production in China were relatively low (Supplementary Fig. 1b). By 2011, five coal bases had exceeded 60 million tons of CO₂ emissions, including YG and SD in western China, the JB and JinZ in central China, and the MD in northeastern China (Supplementary Fig. 1c). The regions with the fastest growth in carbon emissions were primarily concentrated in central and northeastern China. By the end of the study period (Supplementary Fig. 1a and Supplementary Fig. 1d), JD’s emissions surpassed 60 million tons, whereas MD’s emissions decreased, further concentrating coal bases with emissions above 60 million tons in Shanxi Province, central China. Overall, regions with high CO₂ emissions have gradually shifted from the eastern and northeastern areas towards central and western China and are currently concentrated mainly in Shanxi Province. Moreover, the emissions from XJ have continued to increase, whereas emissions in the eastern and northeastern regions have steadily declined.

Policy intensity of green mining policies in China’s coal bases

The annual green mining policy intensity is calculated as the sum of the effective central green mining policy intensity and the green mining policy intensities of the provinces and prefecture-level cities associated with the coal base in that year. To avoid double-counting, if a coal base spans multiple cities within the same province, the provincial green mining policy intensity is counted only once.

The green mining policy intensity in China for 2021, along with the quantified policy intensity metrics for each of the 14 national coal bases from 2004 to 2021, are presented in Fig. 3. The policy intensity of all 14 coal bases tended to increase during the study period (Fig. 3 and Supplementary Fig.3a). In addition, the high-intensity areas affected by the green mining policy exhibited a clear expansion trend (Supplementary Fig. 3d–f). In 2010—the year China issued its national guidelines for advancing green mining development—as a critical policy milestone, only a limited number of coal bases exhibited relatively low policy-intensity values before 2011 (Supplementary Fig. 3d). However, from 2011 onwards, measurable policy intensity emerged across all coal bases (Supplementary Fig. 3e), indicating a nationwide strengthening and diffusion of green mining policy implementation.

Fig. 3: Spatiotemporal map of the intensity of green mining policies in Chinese coal bases, as well as the proportion of intensity and quantity of various types of policy tools and objectives in each coal base.

The Cut-out Shows the South China Sea Islands.

From a regional perspective, the MD in northeastern China recorded both the highest policy intensity and the fastest rate of increase during the study period. Policy intensity first appeared in 2008, increased steadily between 2008 and 2009, increased rapidly from 2009 to 2013, expanded at a slower pace between 2013 and 2016, and accelerated again after 2016, eventually reaching a peak value of 762.67. In eastern and central China, the seven coal bases exhibited relatively similar and concentrated policy-intensity trajectories: policy intensity remained largely stable from 2004 to 2009, increased sharply from 2009 to 2013, increased more moderately between 2013 and 2016, and accelerated again after 2016. Notably, the LH in central China experienced the earliest emergence of policy intensity among all coal bases—already in 2004—despite the absence of formal national green mining policies at that time, as green mining-related requirements appeared in local policy documents. In western China, the SD demonstrated relatively high and rapidly increasing policy intensity. The growth was steady from 2004 to 2009, accelerated between 2009 and 2013, stabilized from 2013 to 2016, and then increased rapidly again after 2016. The policy-intensity patterns of the remaining coal bases were largely consistent with those observed in the eastern and central regions. In 2011, policy intensity was highest in the SD, SB, and JiZ, predominantly concentrated in western China and parts of the eastern region. By the end of the study period, the areas with the highest policy intensity had shifted to the SD, MD, and LH, which were located mainly in northeastern and portions of central and western China. Overall, policy intensity across all coal bases was close to zero in the early years (Supplementary Fig. 2d) but increased substantially by the end of the study period, although both the magnitude of growth and the final intensity levels varied across bases. High-intensity policy regions gradually shifted towards northeastern, central, and western China (Supplementary Fig. 3d and Supplementary Fig. 3f), with the most prominent cluster emerging in the Inner Mongolia Autonomous Region, while other regions also experienced continued policy strengthening. The spatial heterogeneity observed across coal bases may stem from varying responsiveness to national green mining policies, the influence of other related policies, and differences in the number of cities included within each coal base.

Impact of green mining policy on carbon emissions from raw coal production in coal bases

We estimated the policy’s impact on carbon emissions using an interrupted time series framework that incorporates a one-period lag to capture delayed policy effects. Five key parameters (β₀-β₄) reveal the dynamics before and after policy implementation.

First, baseline emissions (β₀) were significantly high across all coal bases, underscoring the sector’s structural carbon intensity prior to policy intervention.

Second, the β₁ coefficients, which represent trends before policy implementation, were generally positive, indicating an underlying upwards trajectory in emissions consistent with expanding raw coal output.

Third, the immediate effect (β₂) of policy introduction was mixed. While a few bases (e.g., SB and HN) exhibited significant negative β₂ values, which suggest an abrupt decline in emissions after policy implementation, most showed statistically insignificant changes, indicating weak short-term policy responsiveness or transition delays.

Fourth, the trend effect (β₃) after policy implementation was consistently negative across all bases, suggesting that green mining policies exert a sustained and time-accumulated suppressive effect on carbon emissions, even when immediate changes are absent.

Fifth, the effect of policy intensity (β₄) was significant only in a subset of bases, namely, SD, SB, and HN. This implies that policy strength alone does not guarantee effectiveness; institutional capacity, policy level, and industrial readiness mediate the actual outcomes. The implementation of the green mining policy has markedly altered the trajectories of raw coal production-related carbon emissions across China’s major coal bases (Fig. 4b–o). Following the implementation of the policy, most coal bases exhibited a clear deceleration or decline in carbon emissions, although the magnitude and timing of these responses varied substantially across regions. From the perspective of coal bases, the coefficients for each base can be broadly classified into four categories: high-response bases, lagging bases, resistance bases and hybrid bases (Fig. 4). High-response bases are those that experienced an immediate decline in carbon emissions following policy implementation and exhibited a sustained long-term downward trend. These include the SB, HN, XJ, LX, and JiZ. In contrast, the lagging bases such as SD, HL, MD, JB, ND, LH, JinZ, JD, and YG did not significantly decrease in the early stage of policy implementation but did decline steadily over the medium and long term. The remaining two categories do not include coal bases. From a regional perspective, coal bases in the eastern region (e.g., LX and JiZ) demonstrated the most rapid and stable responses, followed by those in the western region (e.g., SB and XJ), whereas bases in the central and northeastern regions generally exhibited notable time-lag effects. These patterns suggest that regional differences in policy responsiveness are closely linked to variations in local governance capacity, technological maturity, and industrial structure.

Fig. 4: Impact map of green mining in Chinese coal bases.

a Classification chart of policy impacts on 14 coal bases. b–o Trend charts of carbon emissions from raw coal production in 14 coal bases from 2004 to 2021.

In the high-response bases, all the coefficients for SB and HN are significantly negative (Supplementary Fig. 4k and Supplementary Fig. 4h), indicating that carbon emissions declined immediately after policy implementation and continued to decrease as policy intensity increased. XJ and LX also show an immediate downwards response, although with weaker significance (Supplementary Fig. 4m and Supplementary Fig. 4c). The JiZ has no significant short-term effect but has a clear long-term declining trend (Supplementary Fig. 4b).

In the lagging bases, the immediate effect coefficient for the SD is positive (Supplementary Fig. 4i), suggesting that carbon emissions increased at the early stage of policy implementation, which may reflect the initial phase of closing small mines and shifting production towards larger bases. As the policy advanced, however, carbon emissions significantly and sustainably decreased. HL and MD did not exhibit significant immediate decreases, but their long-term trends were significantly negative (Supplementary Fig. 4l and Supplementary Fig. 4a). The JB, ND, and LH show no significant policy intensity effect (Supplementary Fig. 4d, Supplementary Fig. 4n and Supplementary Fig. 4g), although they still display an initial increase followed by a long-term decline. JinZ, JD, and YG show no significant immediate effects or policy-intensity effects (Supplementary Fig. 4e, Supplementary Fig. 4f and Supplementary Fig. 4j), but carbon emissions generally exhibit a continuous downward trend after policy implementation.

Overall, the green mining policy has effectively reversed the upwards trajectory of carbon emissions from raw coal production in most coal bases, although clear regional heterogeneity and temporal lag effects were evident. The eastern region responds more rapidly and achieves more pronounced reductions, while the central and western regions show more delayed responses. This finding indicates that differences in policy design and implementation capacity and in the technological foundations of the industry jointly shape the variability in emission-reduction outcomes.

Discussion

China’s green mining policy has been implemented for 15 years, with a focus on resource conservation, efficient utilization, pollution control, and ecological restoration in mining areas. This study investigates whether the green mining policy contributes to carbon emission reduction in coal bases and supports the achievement of China’s “dual carbon” goals. Our findings show that between 2004 and 2021, carbon emissions from raw coal production in China’s coal bases continued to increase. Since the implementation of the green mining policy, its intensity has rapidly strengthened to foster resource efficiency, technological advancement, environmental sustainability, and community harmony in mining areas. The results demonstrate a significant long-term positive effect of the policy on carbon reduction in raw coal production, which is particularly pronounced in the central and eastern coal bases. It is estimated that the implementation of the green mining policy has reduced the total carbon emissions from raw coal production in China’s coal bases by 43.59 million tons annually.

From the perspective of regional distribution and coal base characteristics, the MD in Northeast China exhibited a relatively weak immediate effect after the implementation of the policy in 2011. This may be attributed to Inner Mongolia’s earlier actions, as the region initiated resource consolidation and the elimination of outdated production capacity in 2007, well before the national green mining policy. In addition, the promotion of large-scale mineral extraction by key coal enterprises in Inner Mongolia in 2014 further reduced the marginal impact of the national policy introduced in 2011. In the eastern region, the LX shows a similarly weak immediate effect after 2011. This is likely related to Shandong Province’s target in the Twelfth Five-Year Plan, which aimed to increase the planned coal output to 158 million tons by the end of the planning period. This target stimulated rapid capacity expansion within the region and partially offset the carbon reduction effect of the green mining policy. JiZ also has a weak immediate effect and no significant influence of policy intensity on carbon emissions. This may be associated with the severe air pollution in Hebei Province during that period. Since 2012, the province has implemented extensive ecological and environmental governance and mine rectification measures, with the strongest efforts occurring in the early phase. As a result, the marginal benefits of additional policy intensity became relatively small, leading to a statistically insignificant effect of policy intensity on carbon emissions. In the central region, the green mining policy produces substantial mitigation effects in the HN, which may be closely linked to its focus on energy conservation, waste reduction, and resource recycling in mineral resource development. In contrast, the JinZ and JD exhibit weak immediate effects after policy implementation. This could be related to Shanxi Province’s efforts beginning in 2007 to regulate mineral resource development, followed by another round of mine rectification in 2013. Since earlier interventions had already improved resource use efficiency, the additional effect of the national green mining policy was relatively limited. The policy intensity variable is not significant for JB, JinZ, or JD, possibly because of Shanxi’s critical role in the national coal supply. Although these bases have advanced their mining technologies, overall production capacity has continued to increase, and the regional implementation of green mining policy has emphasized environmental rehabilitation more than capacity control. Consequently, the emission reduction effect per unit of policy intensity remains weak. With respect to the LH, the lack of a significant effect of policy intensity on carbon emissions may reflect the concentration of policy issuance within a short period and delayed responses from enterprises.

In the western region, the SD and SB exhibit strong policy responses, with pronounced emission reduction effects following policy implementation. This can be attributed to early phase efforts in both areas to phase out outdated production capacity and promote enterprise consolidation, which facilitated large scale production, supported technological progress, and reduced carbon emissions per unit of output. HL shows a relatively weak immediate effect after the policy was introduced, which may be related to Shaanxi Province’s emphasis since 2012 on energy conservation, emission reduction, and comprehensive resource utilization. These measures led to a substantial decline in production capacity and created a temporal lag in the policy response. XJ also has a weak immediate effect after policy implementation, which may reflect its role as a critical energy replacement and strategic reserve area in the 21st century⁴³. Throughout the study period, the production capacity in Xinjiang continued to expand. Moreover, the region maintained a high level of mining technology and production scale, which reduced the short-term effectiveness of the green mining policy in lowering carbon emissions. The policy intensity in the ND has no significant effect on carbon emissions, which may be related to the concentration of policy issuance combined with relatively delayed enterprise responses. In YG, the immediate effect after the 2011 policy implementation is weak and the impact of policy intensity on carbon emissions is not significant. This may be because Guizhou and Yunnan Provinces focused their efforts on energy conservation during the Twelfth Five-Year Plan period, while the raw coal production capacity continued to expand, thereby weakening the emission reduction effect of the green mining policy.

Overall, the regional differences in response to the green mining policy reflect the complexity of China’s coal industry in terms of policy implementation timing, resource endowment, industrial structure, and governance capacity. The northeastern and central regions generally exhibit delayed policy responses, while the immediate effects are relatively weak in some eastern coal bases because of constraints from production expansion targets. In contrast, the western region shows a stronger emission reduction response, supported by resource integration, large-scale production, and technological progress. This spatial pattern aligns closely with national strategic objectives for energy development and spatial planning, namely compressing production in the east, restricting expansion in the central and northeastern regions, and optimizing development in the west. The distribution of policy effects indicates that green mining policy not only directly contributes to carbon reduction but also exerts structural influence by guiding the spatial reorganization of coal resources and the reallocation of energy production factors. The central and northeastern regions possess substantial resource bases but lag in industrial transformation, so the policy mainly constrains unregulated expansion and promotes technological upgrading. The western region, as a key strategic area for national energy development and reserves, has been promoted by policies to carry out more large-scale, clean, and intensive production.

From the perspective of policy level, central-level policies account for a relatively large share (Supplementary Table 2 and Supplementary Table 3). After the central government formally issued documents on green mine construction, the policy intensity of provincial governments increased, and both the number and intensity of related municipal policies began to grow rapidly. Compared with national policies, provincial policies tailored to local conditions are more targeted; thus, several coal bases had already implemented certain clean production practices to some extent even before the official release of the national green mining policy. But as coal bases have transitioned to large-scale mining operations and have continuously upgraded their technologies, the critical role of coal as a primary energy source necessitates ensuring energy security in China. Carbon emissions from raw coal production in coal bases are likely to stabilize after being reduced to a certain level.

From the perspective of policy instruments, composite policies constitute the majority in terms of both number and intensity across all policies, accounting for 54.74% of the total number of policies and 51.85% of the total policy intensity. Among the various coal bases, composite policies also dominate in both number and intensity, followed by command-and-control policies, whereas market-based policies are the least prevalent and markedly lag behind the other two types. These findings indicate that government or administrative mandates still play a dominant role in carbon emission reduction in China’s coal mining sector, with market mechanisms and social capital participation remaining in the developmental stage. These findings align with previous research highlighting limited policy implementation entities and targets, as well as insufficient incentive mechanisms⁴⁴. Government or administrative mandates may promote carbon reduction in coal mining enterprises by enforcing strict penalties and rectification measures to enhance clean production capacity, whereas market-based incentives primarily encourage enterprises to voluntarily adopt clean production practices through subsidies and rewards. Future measures may be needed to encourage coal mining enterprises to independently and actively engage in green mine construction and to strengthen interdepartmental coordination.

From the perspective of policy objectives, the analyses indicate that policies without specific targets overwhelmingly dominate in both number and intensity across all policies, accounting for 97.01% of the total number and 85.00% of the total intensity. At each coal base, policies without specific objectives markedly outnumber those with defined targets. Policies incorporating all three specific objectives are relatively rare, and those with one or two specific objectives are even rarer. This suggests that while considerable progress has been made in green mines, the specific requirements and objectives still need to be refined and enhanced, which is consistent with prior findings indicating that several policies lack substantive content. Future efforts should refine and clarify policy objectives to better facilitate the implementation and advancement of green mine construction in China.

From the perspective of policy mechanisms, the core pathway through which the green mining policy promotes carbon reduction lies on conservation and efficient utilization. And conservation and efficient utilization is realized through two main channels. First, capacity structure optimization is facilitated by phasing out outdated production capacity and promoting mergers and reorganizations, thereby reducing the number of low-efficiency and high-emission coal mines and generating substantial short-term reductions in carbon emissions. Second, technological improvement in mining operations contributes to long-term declines in carbon intensity by enhancing overall energy efficiency, although its effect appears with a certain lag. In addition, pollution control and ecological restoration objectives indirectly influence carbon reduction. By strengthening the recovery and utilization of associated gases and waste gases and encouraging the adoption of carbon capture, utilization, and storage technologies, green mining policies promote improved environmental management while enhancing carbon sink capacity and supporting the restoration of mining ecosystems.

Considering the differences in policy intensity and operational mechanisms across administrative levels, this study proposes several recommendations for improving green mining policy. At the policy level, the central government should further clarify policy goals and priorities to enhance strategic coordination, while local governments should strengthen policy responsiveness and implementation to ensure effective policy execution. At the policy objective, more detailed targets should be formulated within the existing framework, particularly those promoting carbon capture, utilization and storage deployment and ecological restoration projects from the perspective of increasing carbon sinks, so as to achieve synergistic benefits for carbon reduction and ecological recovery. At the policy instrument, enhancing the initiative of mining enterprises, improving carbon market development, and encouraging the participation of social capital and technological innovation actors in promoting green mine construction are essential.

In conclusion, this study provides empirical evidence regarding the impact of China’s green mining policies on carbon emissions from raw coal production. The results show that these policies have made substantial contributions to reducing carbon emissions in major coal bases. By emphasizing technological upgrading, resource efficiency, and ecological restoration, the design and implementation of green mining policies have effectively advanced the low-carbon transition of the coal sector. Our results contribute to the literature on climate change and mining operations, which typically evaluates carbon emissions and characteristics at individual mine sites while overlooking the independent influence of mining policies, particularly in China. Although the conclusions of this study are drawn from the Chinese context, the mechanisms through which policy interventions influence carbon emissions in the coal mining sector have broader international relevance. Other major coal-producing countries, including Australia, the United States, India, and South Africa, have introduced similar policy frameworks, yet their effects vary. China’s green mining policy has established a comprehensive, government-led system featuring clear objectives and wide coverage, which simultaneously promotes improvements in resource efficiency, technological progress, and ecological restoration, thereby achieving substantial and sustained reductions in carbon emissions from coal production. These findings demonstrate that the effectiveness of carbon reduction policies in the mining sector depends not only on the existence of the policies themselves but also on the degree of institutional coordination, implementation capacity, and integration of multiple governance goals. Therefore, while the quantitative results of this study are based on China, the insights they provide have broader applicability. A policy system characterized by strong institutional coherence and effective implementation mechanisms can generate substantial emission reduction outcomes in carbon-intensive extractive industries. These findings offer important institutional lessons for other coal-dependent economies in balancing resource utilization and environmental protection, advancing efficient and clean energy extraction, and achieving sustainable development objectives.

Methods

To analyze the impact of green mining policy intensity on carbon emissions from raw coal production in coal bases, we first compiled a dataset of carbon emissions from raw coal production in cities associated with green mines. This dataset was combined with raw coal production data from these cities to derive annual carbon emissions for each city. These values were then aggregated to calculate carbon emissions for each coal base, encompassing all relevant cities and regions. Second, we collected data on green mine-related policies at the national, provincial, and municipal levels and quantified their intensity annually for each coal base using a text-based policy analysis method. Finally, we employed the ITSA to examine the effects of green mining policy implementation and its intensity on carbon emissions from raw coal production in coal bases.

The study focuses on 14 coal production bases in China, namely, the SD, SB, HL, JB, JinZ, JD, MD, LH, LX, HN, JiZ, YG, ND and XJ, involving 81 cities, autonomous prefectures or regions across 14 provinces in China (Fig. 5b and Supplementary Table 1). In 2021, China’s total raw coal production reached 4123.905 million tons, with these 14 coal bases contributing 3984.424 million tons, accounting for 96.62% of the nation’s annual output. The comprehensive coverage and dominant production share of these bases make them highly representative for analyzing China’s coal mining sector and its environmental impacts.

Fig. 5: Distribution and production of coal bases in China.

a Raw coal production of 14 coal bases in China from 2004 to 2021. b Distribution map of 14 coal bases in China; the cut-out shows the South China Sea Islands.

Calculation of carbon emissions from raw coal production

The raw coal production process involves greenhouse gas emissions from fuel consumption, net purchased electricity, and the escape of CH₄ and CO₂. Current carbon emission calculation methods include top-down approaches such as the emission factor method, as well as bottom-up methods such as mass balance and direct measurement techniques. Given that the emission factor method²⁵ is particularly suitable for large-scale urban or regional studies, this method is employed in this paper to indirectly estimate carbon emissions from raw coal production on the basis of production data from each city. The emission factors for raw coal production at individual mines were derived using the latest CML2001 (Jan 2016)⁴⁵ methodology published by the Institute of Environmental Sciences at Leiden University, Netherlands (http://energy.qibebt.ac.cn/coalKaicai). The emission factors for raw coal production in each city were then estimated on the basis of the emission factors of individual mines.

1. Determination of the emission factors for raw coal production

(1) Carbon emission coefficient of raw coal production in each mine

The coal mine data used to calculate carbon emission factors were sourced from the Global Energy Monitor website (https://globalenergymonitor.org/projects/global-coal-mine-tracker/). First, on the basis of the classification of China’s coal mines into superlarge, large, medium, and small categories, a random sampling method was applied to select no less than 6% of the mines from each category. For these selected mines, accurate data on coal mining methods, gas grades, and other relevant parameters were collected and analyzed over multiple years. Finally, the carbon emission coefficient for raw coal production was determined using the latest CML2001 methodology⁴⁵, as published by the Institute of Environmental Sciences at Leiden University, Netherlands.

(2) Carbon emission coefficient of raw coal production in each city

Using provinces, coal mining methods, and the coal mine scale as screening criteria, mine data meeting all three standards within each province were obtained. The average carbon emission coefficients for raw coal production in 2021, 2013, and 2004, derived from random sampling, were used as representative values to determine the carbon emission coefficients for mines in these three years. Furthermore, using the carbon emission coefficients of individual mines as benchmarks and cities as retrieval units, data for all mines within each city were collected. The carbon emission coefficients for mines in each city were then calculated on the basis of the proportion of mining quotas. Equation (1) is as follows:

$${{EF}}_{b,c,t}={{E\_coal}}_{b,c,i,t} \cdot {{{{E}_{{EF}}}_{b,c,i,t}/S}_{{coal}}}_{b,c,t},$$

(1)

where ${{EF}}_{b,c,t}$ represents the carbon emission coefficient of raw coal production in city c of coal base b in year t, measured in million tons of CO₂ equivalent per million tons; ${{E\_coal}}_{b,c,i,t}$ represents the production scale of the i mine in city c of coal base b in year t, measured in million tons; ${{E\_EF}}_{b,c,i,t}$ represents the carbon emission coefficient of the i mine in city c of coal base b in year t, measured in million tons of CO₂e per million tons; and ${{S\_coal}}_{b,c,t}$ represents the total raw coal production scale in city c of coal base b in year t, measured in million tons.

Using the three-phase data as a baseline, the carbon emission coefficients for 2004–2021 were proportionally allocated on the basis of the assumption of a steady annual progression in the carbon emission coefficients of raw coal production.

2. Calculation of carbon emissions from raw coal production in each coal base

The annual carbon emissions from raw coal production for each city during the study period were calculated using the method described in 1. (2) The carbon emission coefficient of raw coal production in each city is combined with the raw coal production data for each city. These values were then aggregated to derive the carbon emissions from raw coal production for each coal base, encompassing all relevant prefecture-level cities. Equation (2) is as follows:

$${{Carbon\_emi}}_{b,c,t}={{Coal}}_{b,c,t}\cdot {{EF}}_{b,c,t},$$

(2)

where ${{Carbon\_emi}}_{b,c,t}$ represents the CO₂ emissions from raw coal production in city c of coal base b in year t, measured in 10,000 tons; ${{Coal}}_{b,c,t}$ is the raw coal output in city c of coal base b in year t, measured in 10000 tons; and ${{EF}}_{b,c,t}$ is the carbon emission coefficient of raw coal production in year t in city c of coal base b, measured in 10,000 tons.

$${{Carbon\_emi}}_{b,t}={\sum }_{c=1}^{n}{{Carbon\_emi}}_{b,c,t},$$

(3)

where ${{Carbon\_emi}}_{b,t}$ represents the CO₂ emission from raw coal production of coal base b in year t, measured in 10,000 tons; n is the number of cities involved in coal base b; and ${{Carbon\_emi}}_{b,c,t}$ represents the CO₂ emission in year t in city c of coal base b, measured in 10,000 tons.

Quantification of green mining policy intensity

Policy intensity serves as a key indicator reflecting the significance and enforcement rigor of policies and is quantified through policy objectives and policy instruments^46,47. Policy objectives focus primarily on the formal issues addressed by the policy, whereas policy instruments emphasize the general norms guiding implementation preferences⁴⁸. On the basis of the “policy objective‒policy instrument” framework, this study builds a policy intensity model that integrates policy effectiveness, policy objectives, and policy instruments, considering the contributions of both national and subnational regional policy effects⁴⁹.

1. Data preparation phase

(1) Low-carbon policy data, which include national, provincial, and municipal policy data, were obtained from the PKULAW database (https://pkulaw.com/). Policies were screened on the basis of titles and text containing the term “green mining”, with a focus on their objectives and enforceability, to identify government documents relevant to the construction and management of green mines.

(2) The text of government documents was refined by removing background information, major descriptions, and related introductions. A user-defined lexicon was extracted by analyzing more than five policies, and a reasonable stop-word list was established. The policy objectives were consolidated into three main categories on the basis of the core principles of green mines: (1) exploitation and utilization of mineral resources, (2) mine pollution prevention and ecological restoration, and (3) enterprise management and community harmony. Policy instruments were initially classified into three types: command-and-control, market-based, and composite.

2. Policy text classification

(1) Word frequency statistics of policy texts

The segmented policy texts were analyzed for similarity on the basis of their relevance to the predefined policy objectives and policy instrument keywords. Using a comprehensive lexicon constructed in step 1, which includes keywords from both titles and content, the keywords were identified through an extensive review of the literature. These keywords encompass 28 policy objectives and 21 policy tools. The semantic similarity between the annotated policy content and the keywords was extracted via word2vec, and the synonyms with the closest semantic relevance to the keywords were selected. Terms inconsistent with green mine construction were reviewed and excluded, resulting in the creation of a policy classification dictionary. The policy texts were then classified on the basis of the frequency of relevant phrases.

(2) Classification of green mining policies

For the classification of policy objectives, given that green mining construction is the research focus of this study and that green mining encompass six specific dimensions, these objectives were consolidated into three main objectives. Policies containing all three objectives were classified as “All”, those containing two objectives as “Two”, those containing one objective as “One”, and those containing none as “None”.

For the classification of policy instruments, three categories were defined on the basis of the number of administrative orders and market-based regulations in the policy texts: command-and-control, market-based, and composite. If both command-and-control and market-based instruments were present, the policy was classified as command-and-control if the number of command-and-control instruments exceeded market-based instruments by more than 10 and as market-based if the opposite was true. If the difference was less than 10, the policy was classified as composite. If either command-and-control or market-based instruments were entirely absent, the policy was automatically classified as the other type.

3. Quantification of low-carbon policy intensity

The policy level was integrated into the “policy objective‒policy instrument” framework, and the low-carbon policy intensity for each policy was calculated multiplicatively. Here, ${{PI}}_{r,a,t}$ represents the low-carbon policy intensity of policy $a$ in region $r$ and year $t$, where region $r$ can be a nation ($n$), a province ($\rho$), or a city ($c$). ${L}_{r,a,t,l}$ denotes the intensity of policy level $l$ for policy $a$ in region $r$ and year $t$. ${O}_{r,a,t,o}$ signifies the intensity of policy objective $o$ for policy $a$ in region $r$ and year $t$. ${I}_{r,a,t,i}$ represents the intensity of policy instrument $i$ for policy $a$ in region $r$ and year $t$.

The policy level l encompasses three tiers: national, provincial, and municipal. The policy objectives are classified into four levels: those containing all three specific objectives (All_group), those containing two specific objectives (Two_group), those containing one specific objective (One_group), and those containing no specific objectives (None_group). These objectives are evaluated using a four-level scoring system, which is then normalized to a scale within 3. Policy instruments are categorized into three types: command-and-control (CC), market-based (MB), and composite (CM). The formulas for calculating the policy intensity of region $r$ in year $t$ from the perspectives of policy level, policy objectives, and policy instruments are as follows:

$${{PI}}_{r,t}={\sum }_{a}{{PI}}_{r,a,t},$$

(4)

$${{PI}}_{r,t}={{PI\_All}}_{r,t}+{{PI\_Two}}_{r,t}+{{PI\_One}}_{r,t}+{{PI\_None}}_{r,t},$$

(5)

$${{PI}}_{r,t}={{PI\_CC}}_{r,t}+{{PI\_MB}}_{r,t}+{{PI\_CM}}_{r,t},$$

(6)

in Eq. (4), ${{PI}}_{r,t}$ represents the sum of the policy intensities issued in region $r$ during year $t$. From the perspective of policy levels, ${{PI}}_{r,t}$ can be decomposed into ${{PI\_N}}_{t}$, ${{PI\_P}}_{\rho ,t}$, and ${{PI\_C}}_{c,t}$. Here, ${{PI\_N}}_{t}$ denotes the sum of low-carbon policy intensities at the national level in year $t$, ${{PI\_P}}_{\rho ,t}$ represents the sum of provincial low-carbon policy intensities issued by province $\rho$ in year $t$, and ${{PI\_C}}_{c,t}$ signifies the sum of municipal low-carbon policy intensities issued by prefecture-level city $c$ in year $t$.

In Eq. (5), ${{PI}}_{r,t}$ is the sum of the intensities of the four policy objectives. Specifically, ${{PI}\_{All}}_{r,t}$, ${{PI}\_{Two}}_{r,t}$, ${{PI}\_{One}}_{r,t}$, and ${{PI}\_{None}}_{r,t}$ represent the policy intensities in region $r$ during year $t$ for policies containing all specific objectives, two specific objectives, one specific objective, and no specific objectives, respectively.

In Eq. (6), ${{PI}}_{r,t}$ is the sum of the intensities of the three policy instruments. Here, ${{PI}\_{CC}}_{r,t}$, ${{PI}\_{MB}}_{r,t}$, and ${{PI}\_{CM}}_{r,t}$ denote the intensities of command-and-control, market-based, and composite policies, respectively, in region $r$ during year $t$.

4. Manual labeling

Policy levels, policy objectives, and policy instruments were each scored on a scale ranging from 1 to 3, with a label of 1 indicating weak intensity, a label of 2 indicating moderate intensity, and a label of 3 indicating strong intensity. For policy objectives only, label 4 was used to denote the highest intensity.

(1) In terms of policy objectives, on the basis of word frequency analysis, policies containing all three objectives were classified as the highest level (4), those containing two objectives as level 3, those containing one objective as level 2, and those containing no specific objectives as level 1. The policy objectives were evaluated using a four-level scoring system, which was then normalized to a scale within 3.

(2) Policy instruments were prioritized on the basis of their characteristics. For command-and-control instruments, those specifying exact timelines, strict penalties, bans (e.g., shutdowns, delistings), and official accountability assessments were classified as high intensity (3). Those requiring rectification within a set period or lighter penalties such as downgrading were classified as medium intensity (2), and those with vague or ambiguous language were classified as low intensity (1). For market-based instruments, intensity was determined by the clarity of economic support specified in the policy. For composite instruments, scores were assigned separately for both command-and-control and market-based components, with the maximum of the two scores representing the intensity of the composite instrument.

5. Integration of low-carbon policy intensity

Specifically, when a coal base encompasses only one prefecture-level city, the total green mining policy intensity ${{PI}\_{all}}_{c,t}$ derived from Eq. (7) includes not only the policy intensity at the municipal level but also the policy intensities at higher levels (i.e., national and provincial policies applicable to the city). If a coal base spans multiple provinces or prefecture-level cities, the total policy intensity ${{PI}\_{all}}_{c,t}$ in the Eq. (8) encompasses national policies, as well as relevant provincial and municipal policies. Therefore, the impact of policy intensity on carbon emissions from raw coal production in coal bases must be assessed on the basis of ${{PI}\_{all}}_{c,t}$.

$${{PI\_all}}_{c,t}={{PI\_C}}_{c,t}+{{PI\_aboveCity}}_{c,t}={{PI\_C}}_{c,t}+{{PI\_P}}_{\rho ,t}+{{PI\_N}}_{t},$$

(7)

Policy intervention methods

• 1. The ITSA

  The ITSA is a quasi-experimental design-based analytical method that primarily controls for pre-intervention regression trends and compares immediate changes and trend slope variations before and after the policy intervention point, thereby assessing the impact and effectiveness of the intervention⁵⁰. Initially, applied widely to evaluate the effects of public health policy interventions⁵¹, the ITSA has increasingly been used to other domains, including transportation and land use, among other public policy research areas^52,53.

  The ITSA more effectively controls for pre-existing upwards or downwards trends in the dependent variable prior to policy implementation, thereby providing a more accurate estimation of the policy intervention effect. This approach is particularly effective for analyzing the impact of nationwide uniform policies and significantly mitigates issues related to omitted variable bias. When ITSA is applied, a single-group ITSA is used if no control group is available, whereas a multiple-group ITSA is typically employed when one or more control groups are present⁵⁰.

  In this study, we adopted a single-group design and established a linear regression equation by incorporating policy dummy variables. This enabled a regression analysis of the dependent variables before and after policy implementation and estimation of the intercepts and slopes before and after the interruption point. The ITSA model, which accounts for the impact of the policy and its intensity on carbon emissions, is presented as follows in Eq. (8):

  $${Y}_{t}={\beta }_{0}+{\beta }_{1}{T}_{t}+{\beta }_{2}{D}_{t}+{\beta }_{3}{D}_{t}\left({time}-{T}_{t}\right)+{\beta }_{4}{{policy}}_{t}+\varepsilon,$$

  (8)

  where the dependent variable ${Y}_{t}$ represents the initial value at time point $t$ (i.e., the benchmark level); ${T}_{t}$ can be regarded as a continuous time variable, representing each time point from the beginning of the time series to the implementation of the policy; ${D}_{t}$ is a policy dummy variable, which is 0 before the implementation of the policy and 1 after the implementation; and ${time}$ is another continuous time variable, representing the time points after the time point ${t}^{{\prime} }$ when the policy was implemented. The value of each time point before the implementation is 0, and the value of each time point after the implementation is $\left(t-{t}^{{\prime} }+1\right)$; ${{policy}}_{t}$ indicates the policy intensity in year $T$ after the implementation of the policy.

  The parameter ${\beta }_{0}$ represents the initial intercept, indicating the carbon emissions at the starting point of the time series, which is denoted as $\_{cons}$ in the regression. The parameter ${\beta }_{1}$ indicates the time trend of the dependent variable prior to national policy intervention, representing the annual increase in carbon emissions before policy implementation, typically denoted as $\_t$ in the regression. The parameter ${\beta }_{2}$ captures the immediate effect of the policy intervention, generally denoted as $\_x2011$ in the regression. The parameter ${\beta }_{3}$ reflects the change in the time trend after policy implementation, typically denoted as $\_x\_t2011$ in the regression. The parameter ${\beta }_{4}$ represents the change in carbon emissions resulting from a one-unit change in policy intensity, which is denoted as ${policy}$ in the regression. Finally, $\varepsilon$ is the error term. In the ITSA regression of this study, the primary focus is on the immediate change in carbon emissions (represented by $\_x2011$) and trend changes (represented by $\_x\_t2011$) before and after policy implementation, as well as the effects of unit policy intensity changes (represented by ${policy}$).

• 2. Experimental design

To examine the impact of green mine construction on carbon emissions and its role in advancing energy conservation, emission reduction, and the “dual carbon” goals, carbon emissions from raw coal production in coal bases were selected as the dependent variable. This choice better captures the environmental effects of green mining policies than their administrative impacts.

Considering the representativeness, completeness, and accessibility of the sample data, an annual time series dataset was constructed using green mining policies proposed at the national, provincial, and municipal levels from 2004 to 2021 (18 years in total). Focusing on provinces and cities with substantial coal production within coal bases enables a more accurate reflection and observation of changes in coal mining volume and the optimization of mining methods driven by green mine construction. Green mining policies were sourced from the PKULAW database, and policy intensity was quantified through text-based analysis.

The term “green mining” was first introduced in national-level policy titles in 2010. Therefore, the ITSA model was employed to examine the impact of national green mine construction policy and its intensity on carbon emissions from raw coal production in coal bases following its implementation in 2011. Here, ${\beta }_{2}$ represents the immediate effect of the green mining policy in 2011.

Given that most time series in practice exhibit first-order autocorrelation, failing to correct for autocorrelation may lead to underestimated standard errors or overestimated significance levels of policy effects. To address autocorrelation, the Durbin‒Watson test is typically employed to determine its presence. When autocorrelation is detected, it is controlled by incorporating an autocorrelation parameter into the model. Consequently, the ITSA model was corrected using Newey‒West standard errors to mitigate the downwards bias in coefficient standard errors caused by autocorrelation. The Durbin‒Watson statistics and the corrected model estimates are presented in the figure. Overall, after Newey‒West standard errors were applied for correction, the Durbin–Watson test values approached 2, indicating that the corrected model effectively eliminated autocorrelation issues. Therefore, the regression results from the corrected model were acceptable.

Data uncertainty analysis

In this study, coal mine carbon emissions refer to carbon emissions throughout the entire lifecycle of coal mining, including greenhouse gas emissions generated by all production designs and ancillary facilities under operational control. The scope of these facilities includes the primary production system, auxiliary production systems, and ancillary production systems that directly support production. The auxiliary production systems include transportation, hoisting, drainage and dust suppression systems, as well as a power supply, heating, cooling, mechanical repair, and storage within the plant area. The supporting production systems include production command systems (plant headquarters) and departments or units serving production within the plant area (e.g., staff canteens and workshop bathrooms). Specifically, emissions can be categorized into four types: CO₂ emissions from fuel combustion, CH₄ fugitive emissions, net purchased electricity, and water usage. It does not include carbon emissions associated with methane leakage during the coal extraction process. The carbon emission coefficients are compared with existing studies on the lifecycle carbon emissions of coal mines in regions such as the Inner Mongolia Autonomous Region⁵⁴, Shanxi Province⁵⁵, Shaanxi Province, and Gansu Province⁵⁶. For example, the difference between the carbon emission coefficient for a typical open-pit mine in Ordos City, Inner Mongolia, in 2021 and the corresponding coefficient in this study is 0.0109⁵⁴, resulting in a carbon emission coefficient error of 21.8%. Overall, the carbon emission coefficient error was 23.3%.

The method employed in this study utilizes relatively generalized emission factors for fuel combustion (including bituminous coal, natural gas, gasoline, diesel, alcohol-based fuels, explosives, etc.), as well as for fugitive carbon emissions, electricity consumption emissions, and carbon emissions from new water consumption. In contrast, other studies focused on the lifecycle carbon emissions of specific mines provide more accurate and targeted assessments that better reflect the characteristics of the research subjects. Future research could adjust these factors on the basis of specific data from various regions or provinces in China to obtain more precise carbon emission coefficients, thereby increasing the accuracy of future carbon emission estimates.

The methodology employed in this study adopts relatively generalized values for emission factors associated with fuel combustion (including bituminous coal, natural gas, gasoline, diesel, alcohol-based fuels, explosives, etc.), as well as for fugitive carbon emission factors, electricity consumption emission factors, and freshwater consumption carbon emission factors. In contrast, other life cycle carbon emission studies tailored to specific mines offer more refined and targeted assessments that align better with the actual conditions. Future research could be adjusted based on specific information from various regions, provinces, or municipalities in China to derive more precise carbon emission coefficients, thereby enhancing the accuracy of future carbon emission estimates.

Second, due to the absence of data for certain cities, not all cities within some coal bases were included in this study. These cities include Liaoyang City in the MD, Bozhou City in the LH, Heze City and Bozhou City in the LX, Jiyuan City in the HN, Langfang City and Cangzhou City in the JiZ, Yibin City and Luzhou City in the YG, and Wuzhong City in the ND. Moreover, as some cities are affiliated with multiple coal bases, to avoid substantial errors, they were only accounted for within one coal base in our calculations.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.