Climate Impact of China’s Promotion of the Filling Mining Method: Bottom-Up Estimation of Greenhouse Gas Emissions in Underground Metal Mines

: China recently implemented a “Green Mine” policy focused on promoting the ﬁlling method, aiming to mitigate the environmental impacts of underground mining; nevertheless, quantitative inventories have rarely been provided to support or negate such promotion, especially from a life-cycle perspective. Accordingly, this paper proposes a bottom-up model for estimating life-cycle greenhouse gas (GHG) emissions from underground metal mines using either ﬁlling or caving methods. Two ﬁlling-based (Luohe and Longtangyan) and two caving-based (Maogong and Xiaowanggou) iron mines were studied; their direct GHG emissions were 0.576, 0.278, 2.130, and 1.425 tons of carbon dioxide equivalent per kiloton-extracted ore (t CO 2 eq/kt), respectively. When indirect GHG emissions were considered, the results increased to 17.386, 15.211, 5.554, and 5.602 t CO 2 eq/kt, respectively. In contrast to popular belief, such results demonstrate that promoting the ﬁlling method can potentially raise the overall GHG emissions. Although ﬁlling-based projects generate less direct GHG emissions, the emissions are transferred to upstream sectors, especially the cement and power sectors. The additional electricity consumption in the haulage and backﬁlling stages is primarily responsible for the greater GHG emissions occurring in ﬁlling-based projects. Some mitigation approaches are suggested, such as backﬁlling the subsidence pit, using industrial waste as cementing materials, employing energy-efﬁcient pumps, and further developing hauling systems.


Introduction
China has been the largest greenhouse gas (GHG) emitter for more than 15 years [1]. To mitigate global climate change, China has implemented numerous policies to conserve energy and reduce GHG emissions in different industrial sectors [2][3][4]. "Green Mine" is a mandatory policy recently released by China's Ministry of Natural Resources (MNR), which aims to mitigate the environmental impact of the mining sector. In this policy, the filling method, rather than open-stope or caving methods, is promoted in China's underground mines [5].
In the existing literature, the filling method is regarded as a cleaner method for extracting metal deposits, owing to the preservation of the ground surface and the associated vegetation [6,7]. In contrast, underground caving leads to large-scale surface subsidence [8]; this means that the vegetation on the ground surface will be destroyed. Moreover, the land occupation for storing waste rocks and tailings further deteriorates such destruction. Such drawbacks restrict the application of caving methods in China [9,10]. However, quantitative assessments have rarely been proposed to test the differences in the environmental and backfilling, and surface destruction. Fossil fuel, industrial explosives, electricity, cement, and carbon sink losses constitute the GHG emission sources. The GHG emissions of a mining project can be estimated by aggregating the contributions of each source in the various stages. The functional unit in this LCA analysis is 1 ton of CO2 equivalent per kiloton of excavated ore (t CO2 eq/kt ore) transported to the milling plants. The system boundary enables the modeling of GHG emissions for both caving-based and filling-based metal mines. An inventory analysis is initially conducted for the emission factors of the involved GHG sources.

Fossil Fuel
The IPCC [25] has provided the default emission factors of CO2, CH4, and N2O from the combustion of different fossil fuels. Such data can be normalized to satisfy the requirement of the functional unit, according to the higher heating values (HHVs) [26] and global warming potential (GWP) [27]. The indirect GHG emission factors of fossil fuels can be The functional unit in this LCA analysis is 1 ton of CO 2 equivalent per kiloton of excavated ore (t CO 2 eq/kt ore) transported to the milling plants. The system boundary enables the modeling of GHG emissions for both caving-based and filling-based metal mines. An inventory analysis is initially conducted for the emission factors of the involved GHG sources.

Emission Factors of GHG Emission Sources 2.2.1. Fossil Fuel
The IPCC [25] has provided the default emission factors of CO 2 , CH 4 , and N 2 O from the combustion of different fossil fuels. Such data can be normalized to satisfy the requirement of the functional unit, according to the higher heating values (HHVs) [26] and global warming potential (GWP) [27]. The indirect GHG emission factors of fossil fuels can be derived from the literature regarding the GHG emissions in oil fields and refineries, such as [28,29]. Such emission factors are provided in Table 1. Table 1. Normalized GHG emission factors of fossil fuels.

Fossil Fuel
EF j (t CO 2 eq/t) EF j (t CO 2 eq/t) EF j (t CO 2 eq/t) Notes: (1) EF m is the direct GHG emission factor from the combustion of fossil fuel j; EF j the indirect GHG emission factor for producing the fossil fuel j; EF j is GHG emission factor of fossil fuel j. (2) The direct GHG emission factors are calculated by EF j = ∑ i DEF ij ·HHV j ·GWP i [30]; i represents the type of greenhouse gas, e.g., CO 2 , CH 4 , or N 2 O; j represents the type of fossil fuel, e.g., gasoline or diesel; DEF ij is the default emission factor of greenhouse gas i from fossil fuel j combustion [25]; HHV j is the higher heating value of fossil fuel j [26]; GWP i is the global warming potential of greenhouse gas i [27]. (3) The indirect GHG emission factors are derived from [28,29]. (4) The statistical recovery ratio of light fossil fuel in China's oil field and refinery is 74% [31]. (5) The GHG emissions occurring in the transportation from refineries to mining areas is neglected.

Electricity
Electricity is a clean energy for metal mines. However, GHG will be emitted to generate electricity, especially in China, because coal-fired plants still play a predominant role in its power supply [32]. Table 2 lists the default GHG emission factors of electricity for different regional power systems, provided by China's NDRC [33]. Table 2. Baseline GHG emission factors of China's regional power grid by NDRC.

Industrial Explosives
Industrial explosives are mixtures consisted of a blasting agent (e.g., ammonium nitrate, AN) and additives (e.g., diesel, wood flour, and emulsifier), and the GHG emission factors heavily depend on the share of each component (Table 3). (2) EF n is the direct GHG emission factor in the blasting reaction of industrial explosive n, which is calculated by the B-W approach [35]. (3) EF AN is the indirect GHG emission factor for producing AN [36]. (4) EF Diesel is the indirect GHG emission factor for producing diesel in oilfields and refineries [28,29]. (5) EF n is the GHG emission factor of industrial explosive n.
The Brinkley-Wilson (B-W) approach is one of the most commonly cited approaches to calculate the GHG emission factors of industrial explosives [35]. The indirect GHG emissions of AN and diesel are considered in the proposed model. The indirect GHG emission factor of AN is obtained by referring to the literature regarding AN fertilizer production [36], and the diesel factor is shown in Table 1. Table 3 lists the GHG emission factors of some industrial explosives commonly utilized in China's metal mines.

Cement
Filling-based mines consume a large quantity of cement to develop a stable filling body. Underground tunneling also requires cement to enhance the surrounding rocks, in both filling-based and caving-based projects. Cement is an indirect GHG emission source for underground mines. This approach, validated by the IPCC, enables estimations of the GHG emissions of clinker production [37]. Table 4 lists literature regarding GHG emissions of China's cement sector.

Carbon Sink Loss
A carbon sink of vegetation facilitates the mitigation of global climate change [41,42]. It can be quantitively represented by the net primary production (NPP) [43,44], according to the land use categorized by the IPCC [45], e.g., forestland, cropland, grassland, wetlands, settlements, and other land uses. Table 5 lists the NPPs of the land with different vegetations.

Bottom-Up Model for GHG Emissions
Herein a bottom-up model is proposed to estimate the life-cycle GHG emissions of extracted ore in the metal mines by different methods, by analytically relating the inputs and outputs involved in the LCA analysis. Besides the GHG emission factors listed, we involve engineering-based parameters as inputs, such as property of mining equipment (e.g., power, average consumption for fossil fuel), operating time, and predicted area of surface destruction.

Ventilation
Underground metal mines, both caving-based and filling-based, employ a fan system to bring freshness to working faces. GHG emissions ocurr due to the consumption of electricity by the fan system, which can be calculated by Equation (1): where E i 1 is the annual GHG emissions in the ventilation stage in the underground mine by method i; EF ele and EF lub are the GHG emission factors of electricity and lubricating oil, respectively; P 1,m is the power of the ventilation equipment m, e.g., the fan system or the local fans; Q lub 1,m is the quantity of lubricating oil consumed by ventilation equipment m in unit time; t 1,m is the annual operating time of ventilation equipment m.

Drilling
Drilling is required to make blasting holes for tunneling or ore excavating. Electricitypowered drills are preferred in underground metal mines to maintain the environment in working faces, but different designs and equipment are involved for different purposes. The GHG emissions due to electricity and lubricating oil consumed in the drilling stage can be estimated by Equation (2): where E i 2 is the annual GHG emissions in the drilling stage in the underground mine by method i; P 2,m is the power of the underground drill m; Q lub 2,m is the quantity of lubricating oil consumed by the underground drills m in unit time; t 2,m is the annual operating time of underground drill m.

Blasting
Industrial explosives are used in underground mines to break intact rock. Different industrial explosives and associated designs are implemented when tunneling and excavating ore. According to the consumption of industrial explosives, the GHG emissions in the blasting stage can be estimated by Equation (3): where E i 3 is the annual GHG emissions in the blasting stage in the underground mine by method i; EF exp m is the GHG emission factor of industrial explosive m; Q exp 3,m is the annual consumption of industrial explosive m.

Loading
Loading equipment, such as the load-haul-dump (LDH) and the scraper, is utilized to collect and load broken ore into the hauling equipment. Because the LDH integrates ore collecting and hauling, its GHG emissions are calculated in the hauling stage. In addition, vibrating feeders are also deployed between different equipment or operations, such as at the exit of the draw shaft or in the ore transfer station. Equation (4) is proposed to estimate the GHG emissions in this stage: where E i 4 is the annual GHG emissions in the loading stage in the underground mine by method i; P 4,m is the power of electricity-powered loading equipment m; Q lub 4,m is the quantity of the lubricating oil consumed by loading equipment m in unit time; t 4,m is the annual operating time of loading equipment m.

Haulage
The haulage system of an underground mine can be categorized as working face hauling, underground hauling, underground-to-surface hoisting, and surface haulage. The load-haul-dump (LDH) is the most commonly utilized hauling equipment in the working face of both caving-based and filling-based underground mines. Trucks, electric locomotives, and belt conveyers are mainstream equipment in underground and surface haulage. Skip buckets or cages are utilized to lift the ore or waste rock from underground to surface. Even though electricity-powered equipment is preferred, diesel-powered equipment is sometimes involved. The GHG emissions due to fossil fuel, electricity, and lubricating oil consumed in the haulage stage can be predicted by Equation (5): where E i 5 is the annual GHG emissions in the haulage stage in the underground mine by method i; EF die is the GHG emission factor of the diesel; P 5,m is the power of electricitypowered hauling equipment m, e.g., electricity-powered LDHs, electric locomotives, belt conveyers, and skip buckets; Q die 5,n is the quantity of diesel consumed by diesel-powered hauling equipment n in unit time, e.g., diesel-powered LDHs and trucks; Q lube 5,m&n is the quantity of lubricating oil consumed by hauling equipment m or n in unit time; t ele 5,m and t die 5,n are the annual operating times of either the electricity-powered or the diesel-powered hauling equipment, respectively; t 5,m&n is the annual operating time of hauling equipment m or n; ∑ m t ele 5,m + ∑ n t ele 5,m = ∑ m+n t 5,m+n .

Crushing
Due to the requirement of the block size by the hauling or loading equipment, crushing equipment is deployed in underground mines. The GHG emissions in the crushing stage primarily come from electricity consumption, which can be calculated by Equation (6): where E i 6 is the annual GHG emissions in the crushing stage in the underground mine by method i; P 6,m is the power of crusher m, e.g., jaw crusher or gyratory crusher; Q lub 6,m is the quantity of lubricating oil consumed by crusher m in unit time; t 6,m is the annual operating time of crusher m.

Support and Backfilling
Cement is used in both caving-based and filling-based metal mines to support the rock around tunnels and accesses. Filling-based projects require additional cement to enhance the stability of the filling bodies. Cement use in metal mines leads to indirect GHG emissions. Electricity is also consumed in the support and backfilling stage to power the associated equipment, such as a blender and mortar pump. Equation (7) can estimate GHG emissions in this stage: where E i 7 is the annual GHG emissions in the support and backfilling stage in the underground mine by method i; EF cem is the GHG emission factor of the cement; Q cem 7,m is the annual consumption of cement for purpose m; e.g., supporting tunnels or backfilling underground void; P 7,m is the power of the equipment m involved in the support and backfilling stage; t 7,m is the annual operating time of equipment m.

Surface Destruction
In caving-based metal mines, underground excavation leads to large-scale surface subsidence. Some robust approaches have been proposed to predict the boundary of the subsidence area [48][49][50][51]. The land occupation for dumping waste rocks and tailings, which takes place in both caving-based and filling-based projects, can be obtained from design or in situ investigation. The GHG emissions due to carbon sink loss can be estimated by Equation (8): where E i 8 is the annual GHG emissions due to carbon sink variation in the underground mine by method i; α is the conversion coefficient from net primary production to GHG emissions; NPP 1 and NPP 2 are the net primary production of the vegetation on the original and eventual ground surface, respectively; S cs is the accumulative area of land whose use changed due to underground mining.
Therefore, the GHG emissions in an underground metal mine employing either the caving or the filling method can be estimated by accumulating the results by Equations (1) to (8): where E i is the GHG emission in an underground metal mine by method i; E i j is the GHG emissions in the mining stage j involved in the underground metal mine by method i. The proposed model can obtain detailed emission data from each GHG source in varying mining stages. This means that it contributes to uncovering differences in climate impact due to employment of either the filling or the caving method. The unit GHG emissions thus can be estimated in accordance with the result by Equation (9) and the annual production in a mining project.
Additionally, the limitations of the proposed model should be discussed. This model is valid to estimate the GHG emissions occurred in the operations involved in the system boundary. More operations are also required in the underground metal mines, such as lighting, road construction. It thus can be expected that the GHG emissions take place in-situ is likely to be higher than the predicted by the proposed model.

Inputs
To test the climate impact of China's promotion policy for the filling method, four underground iron mines in China were considered, including two caving-based (i.e., Luohe and Longtangyan) and two caving-based (i.e., Maogong and Xiaowanggou) mines. Basic information regarding these mines is summarized in Table 6. Notes: (1) The location of each project relates to the employment of the GHG emission factor of electricity and the NPP of vegetation on the ground surface. (2) The surface destruction for surface constructions and waste dumps is neglected for all involved cases. (3) The surface subsidence area is determined based on the deposit distribution and the caving angle. Table 6 shows that the production capacity of Luohe is similar to that of Maogong, whereas Longtangyan's capacity is similar to Xiaowanggou's. Notably, both filling and caving methods are applicable for these four projects. The Maogong and Xiaowanggou mines employ the caving method, owing to a preference for a higher cost efficiency [52]. As surface subsidence is forbidden in Luohe and Longtangyan, they utilize the filling method. The sources of the data for estimating GHG emissions should be noted. The emission factors for each GHG source have been provided in Tables 1-5. We collected the designs currently implemented in the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines, so as to determine the equipment involved in each project, as well as the associated engineering-based inputs required by the bottom-up model to estimate the life-cycle GHG Energies 2021, 14, 3273 9 of 17 emissions. It was expected that the engineering-based results from the bottom-up model could explicitly identify the underlying drivers of GHG emissions from the projects with the different mining methods.

Results
The GHG emission factors listed in Tables 1-5 and the inputs derived from the implemented designs in the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines enable the estimation of their GHG emissions. The GHG emissions of each involved case and the associated distributions are illustrated in Figures 2 and 3, respectively.
Energies 2021, 14, x FOR PEER REVIEW 9 of 17 sion factors for each GHG source have been provided in Tables 1-5. We collected the designs currently implemented in the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines, so as to determine the equipment involved in each project, as well as the associated engineering-based inputs required by the bottom-up model to estimate the lifecycle GHG emissions. It was expected that the engineering-based results from the bottomup model could explicitly identify the underlying drivers of GHG emissions from the projects with the different mining methods.

Results
The GHG emission factors listed in Tables 1-5 and the inputs derived from the implemented designs in the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines enable the estimation of their GHG emissions. The GHG emissions of each involved case and the associated distributions are illustrated in Figures 2 and 3, respectively.    Tables 1-5. We collected the designs currently implemented in the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines, so as to determine the equipment involved in each project, as well as the associated engineering-based inputs required by the bottom-up model to estimate the lifecycle GHG emissions. It was expected that the engineering-based results from the bottomup model could explicitly identify the underlying drivers of GHG emissions from the projects with the different mining methods.

Results
The GHG emission factors listed in Tables 1-5 and the inputs derived from the implemented designs in the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines enable the estimation of their GHG emissions. The GHG emissions of each involved case and the associated distributions are illustrated in Figures 2 and 3, respectively.    Figure 2 shows that the GHG emissions for excavating a unit quantity of ore in Luohe (17.386 t CO 2 eq/kt) and Longtangyan (15.211 t CO 2 eq/kt) are much higher than those in Maogong (5.554 t CO 2 eq/kt) and Xiaowanggou (5.602 t CO 2 eq/kt). This indicates that the mining method, rather than the production capacity, is responsible for the differences in the GHG emissions between projects. The GHG emission distributions in cavingbased and filling-based projects, as illustrated in Figure 3, provide additional details regarding these differences. Figure 3 shows that electricity consumption dominates the GHG emissions in both caving-based and filling-based projects, but its share in the fillingbased mines (87.422% and 88.374% in the Luohe and Longtangyan iron mines, respectively) is much higher than in caving-based mines (48.703% and 57.881% in the Maogong and Xiaowanggou iron mines, respectively). Notably, the electricity consumption values are 18.890, 16.708, 2.441, and 2.926 MWh/kt in the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines, respectively, but their associated GHG emissions are 15.199, 13.443, 2.705, and 3.243 t CO 2 eq/kt, respectively. The lower GHG emission factor of the electricity in Anhui (Table 3), where the Luohe and Longtangyan iron mines are located, reduces their GHG emissions. Cement is another contributor to the higher GHG emissions in the filling-based cases; the contributions are 0.926, 0.456, 0.167, and 0.003 t CO 2 eq/kt in the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines, respectively. The Luohe and Longtangyan iron mines require 57.877 and 30.387 t of cement to excavate 1000 t of iron ore, whereas those values are 10.415 and 0.166 t/kt in the Maogong and Xiaowanggou iron mines, respectively. The additional cement required in the filling-based cases is primarily used for filling in voids after deposit excavation. This indicates that using low-carbon cementing materials is a robust approach to mitigating GHG emissions in filling-based projects.
In addition, the carbon sink losses in caving-based projects raises their GHG emissions. These emissions in the Maogong and Xiaowanggou mines are 1.743 and 0.702 t CO 2 eq/kt, respectively, whose shares of the total GHG emissions are 31.385% and 12.529%, respectively. This indicates that surface subsidence control and timely reclamation of subsidence pits can mitigate the GHG emissions in caving-based projects. Notably, only the consumption of industrial explosives and diesel and carbon sink losses lead to direct GHG emissions. Thus, it is expected that the GHG emissions in filling-based projects will be lower than those in caving-based projects if only direct GHG emissions are considered in the estimation. The results of such an estimation are shown in Figure 4.  Figure 2 shows that the GHG emissions for excavating a unit quantity of ore in Luohe (17.386 t CO2 eq/kt) and Longtangyan (15.211 t CO2 eq/kt) are much higher than those in Maogong (5.554 t CO2 eq/kt) and Xiaowanggou (5.602 t CO2 eq/kt). This indicates that the mining method, rather than the production capacity, is responsible for the differences in the GHG emissions between projects. The GHG emission distributions in caving-based and filling-based projects, as illustrated in Figure 3, provide additional details regarding these differences. Figure 3 shows that electricity consumption dominates the GHG emissions in both caving-based and filling-based projects, but its share in the filling-based mines (87.422% and 88.374% in the Luohe and Longtangyan iron mines, respectively) is much higher than in caving-based mines (48.703% and 57.881% in the Maogong and Xiaowanggou iron mines, respectively). Notably, the electricity consumption values are 18.890, 16.708, 2.441, and 2.926 MWh/kt in the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines, respectively, but their associated GHG emissions are 15.199, 13.443, 2.705, and 3.243 t CO2 eq/kt, respectively. The lower GHG emission factor of the electricity in Anhui (Table 3), where the Luohe and Longtangyan iron mines are located, reduces their GHG emissions. Cement is another contributor to the higher GHG emissions in the filling-based cases; the contributions are 0.926, 0.456, 0.167, and 0.003 t CO2 eq/kt in the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines, respectively. The Luohe and Longtangyan iron mines require 57.877 and 30.387 t of cement to excavate 1000 t of iron ore, whereas those values are 10.415 and 0.166 t/kt in the Maogong and Xiaowanggou iron mines, respectively. The additional cement required in the fillingbased cases is primarily used for filling in voids after deposit excavation. This indicates that using low-carbon cementing materials is a robust approach to mitigating GHG emissions in filling-based projects.
In addition, the carbon sink losses in caving-based projects raises their GHG emissions. These emissions in the Maogong and Xiaowanggou mines are 1.743 and 0.702 t CO2 eq/kt, respectively, whose shares of the total GHG emissions are 31.385% and 12.529%, respectively. This indicates that surface subsidence control and timely reclamation of subsidence pits can mitigate the GHG emissions in caving-based projects. Notably, only the consumption of industrial explosives and diesel and carbon sink losses lead to direct GHG emissions. Thus, it is expected that the GHG emissions in filling-based projects will be lower than those in caving-based projects if only direct GHG emissions are considered in the estimation. The results of such an estimation are shown in Figure 4. The direct GHG emissions illustrated in Figure 4 arise from industrial explosive blasting, diesel combustion, and carbon sink losses directly occurring in the mining areas of The direct GHG emissions illustrated in Figure 4 arise from industrial explosive blasting, diesel combustion, and carbon sink losses directly occurring in the mining areas of the studied cases. This means that neither the GHG emissions from indirect sources (e.g., electricity, cement, and lubricating oil) nor the indirect GHG emissions from producing  Figure 4. Figure 4 shows that the direct GHG emissions from caving-based projects (2.130 and 1.425 t CO 2 eq/kt in Maogong and Xiaowanggou, respectively) are higher than those from filling-based projects (0.576 and 0.278 t CO 2 eq/kt in Luohe and Longtangyan, respectively). The results illustrated in Figures 3 and 4 indicate that China's promotion policy regarding the filling method contributes to mitigating the direct GHG emissions occurring in the mining areas; however, this mitigation works by transferring GHG emissions to the suppliers of energy and materials, such as power plants and cement plants.
Considering that filling-based cases require much more electricity than caving-based cases, we analyzed the electricity consumption in each mining stage, and the associated distributions are illustrated in Figures 5 and 6, respectively.
Energies 2021, 14, x FOR PEER REVIEW 11 of the studied cases. This means that neither the GHG emissions from indirect sources (e. electricity, cement, and lubricating oil) nor the indirect GHG emissions from producin industrial explosives and diesel are involved in the results illustrated in Figure 4. Figu  4 shows that the direct GHG emissions from caving-based projects (2.130 and 1.425 t C eq/kt in Maogong and Xiaowanggou, respectively) are higher than those from fillin based projects (0.576 and 0.278 t CO2 eq/kt in Luohe and Longtangyan, respectively). T results illustrated in Figures 3 and 4 indicate that China's promotion policy regarding t filling method contributes to mitigating the direct GHG emissions occurring in the minin areas; however, this mitigation works by transferring GHG emissions to the suppliers energy and materials, such as power plants and cement plants.
Considering that filling-based cases require much more electricity than caving-bas cases, we analyzed the electricity consumption in each mining stage, and the associat distributions are illustrated in Figures 5 and 6, respectively.   Figure 5 shows that the haulage stage and the support and backfilling stage are t primary contributors to the higher electricity consumption in filling-based projects (i. the studied cases. This means that neither the GHG emissions from indirect sources (e.g., electricity, cement, and lubricating oil) nor the indirect GHG emissions from producing industrial explosives and diesel are involved in the results illustrated in Figure 4. Figure  4 shows that the direct GHG emissions from caving-based projects (2.130 and 1.425 t CO2 eq/kt in Maogong and Xiaowanggou, respectively) are higher than those from fillingbased projects (0.576 and 0.278 t CO2 eq/kt in Luohe and Longtangyan, respectively). The results illustrated in Figures 3 and 4 indicate that China's promotion policy regarding the filling method contributes to mitigating the direct GHG emissions occurring in the mining areas; however, this mitigation works by transferring GHG emissions to the suppliers of energy and materials, such as power plants and cement plants.
Considering that filling-based cases require much more electricity than caving-based cases, we analyzed the electricity consumption in each mining stage, and the associated distributions are illustrated in Figures 5 and 6, respectively.   Figure 5 shows that the haulage stage and the support and backfilling stage are the primary contributors to the higher electricity consumption in filling-based projects (i.e., Figure 6. Distribution of electricity consumption in the Luohe, Longtangyan, Maogong, Xiaowanggou iron mines. Figure 5 shows that the haulage stage and the support and backfilling stage are the primary contributors to the higher electricity consumption in filling-based projects (i.e., the Luohe and Longtangyan iron mines) relative to caving-based projects (i.e., the Maogong and Xiaowanggou iron mines). The consumption values of electricity in the haulage stage are 5.409, 3.789, 1.343, and 1.494 kWh/t for the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines, respectively, with shares of the total emissions of 35.587%, 28.188%, 49.657%, and 46.078%, respectively. The higher efficiency that the hauling equipment presents in the caving-based cases accounts for this phenomenon. For instance, the Luohe iron mine has a production capacity similar to that of Maogong, but its larger load-haul-dump machines (LHDs) (type: Sandvik LH514E, bucket volume: 6 m 3 ) present a lower efficiency (400 kt/a) than the smaller LHDs in Maogong (type: CY-4), whose bucket volume and efficiency are 4 m 3 and 675 kt/a, respectively. Coincidentally, the efficiency of the LHDs in the Xiaowanggou iron mine (type: Scooptram EST3.5, bucket volume: 2.3 m 3 , efficiency: 320 kt/a) is also higher than that of the LHDs in Longtangyan (bucket volume: 2 m 3 , hauling capacity: 200 kt/a). The developed hauling system accounts for this phenomenon, for instance, the preparation ratio in Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines are 2.55, 2.33, 8, and 4.43 m/kt. This indicates that a reduction in GHG emissions in filling-based projects can be expected if the efficiency of the hauling equipment can be improved. Moreover, further developing the efficiency of the hauling equipment can also help to mitigate GHG emissions in caving-based projects, as the hauling stage dominates their electricity consumption ( Figure 6). In contrast, the support and backfilling stage constitutes 26.374% and 34.692% of the total electricity consumption in the Luohe and Longtangyan iron mines, respectively ( Figure 6). Such values in the Maogong and Xiaowanggou mines are 6.517% and 0.404% (Figure 6), respectively. The additional consumption of electricity in the filling-based projects is primarily utilized to power pumps to either transport or dewater the filling slurry. Such phenomena indicate that approaches for conserving the energy consumed by pumps can also mitigate GHG emissions in filling-based projects.

Discussion and Implications
The filling method has been promoted by China's Green Mine policy, which was implemented in 2020. Owing to China's target responsibility system, it can be expected that the share of filling-based projects in China's mineral supply will significantly increase in the following decade.
The results from the case studies of Luohe, Longtangyan, Maogong, and Xiaowanggou show that filling-based projects generate less direct GHG emissions than those based on caving, indicating that the aforementioned promotion helps to mitigate GHG emissions that occur directly in the mining sector. However, when indirect GHG emissions are considered, the GHG emissions in filling-based cases present a more notable increase, and exceed the results of the caving-based cases. This reveals that the mitigation of GHG emissions in filling-based projects works by transferring the emissions upstream to cement and power plants. This means that China's promotion policy potentially raises the overall GHG emissions of its domestically extracted mineral products.
Moreover, the GHG emissions owing to the carbon sink losses occurring in the subsidence areas in Maogong and Xiaowanggou are likely to be lower than those in the predictions illustrated in Figure 3, owing to the involvement of certain assumptions in the estimation. First, the surface subsidence, as well as the associated vegetation destruction, is assumed to occur immediately after the underground caving is implemented. However, surface subsidence is a progressive process that lasts for a long period of time (Figure 7). Another assumption is that no carbon sink will occur once surface subsidence has occurred. Although surface subsidence destroys the forest on the ground surface, the recovery of herbaceous plants has been observed in the subsidence area of the Xiaowanggou iron mine ( Figure 5). Therefore, the carbon sink loss prediction illustrated in Figure 3 is greater than the actual carbon sink losses in the Maogong and Xiaowanggou iron mines. not yet suitable for China, especially under a target for a carbon peak before 2030. However, such a promotion can be expected to eventually contribute to mitigating the overall GHG emissions in the mining sector if the GHG emission factor of China's power supply maintains the decreasing trend presented in the past decade. Additionally, further suggestions aimed at mitigating GHG emissions in underground metal mines can be provided. Although existing literature has proposed numerous policies or technologies, such as taxing GHG emissions [53], carbon cap and trading [54], developing efficiency in diesel engines [55], and increasing the share of renewable energy in the power supply [16], the case studies suggest more elaborate mitigation approaches to these projects. For example, backfilling of the subsidence pit is suggested for caving-based cases, so as to prevent the subsequent extension of surface subsidence after vertical caving. Such prevention has been validated in the Hemushan and Gongchanling iron mines [56]. Moreover, three approaches, i.e., using industrial wastes (such as fly ash or alkali-activated slag [57,58]) as cementing materials to reduce the consumption of cement in the backfilling stage, employing energy-efficient pumps, and further developing hauling systems, especially LHDs, are recommended for filling-based projects to mitigate their GHG emissions.
The uncertainty of this study should be clarified. Uncertainty primarily exists in the modelling for GHG emission estimation. The proposed model is a static input-output model, which means all input, e.g., emission factor of GHG sources and mining equipment, are assumed to be constant. However, these parameters are likely to varies along with the employment of GHG-abatement technologies [59] and the advance of power plants and oil refineries. The dynamic analysis, especially considering the potential of underground metal mines to mitigate GHG emissions, can be applied to deal with this uncertainty. Another uncertainty source is the variation of backfilling materials. The fillingbased cases involved (i.e., Luohe and Longtangyan) use cemented materials, and plenty In view of the above, as promoting the filling method will potentially raise the GHG emissions of China's domestically extracted mineral products, mandatory promotion is not yet suitable for China, especially under a target for a carbon peak before 2030. However, such a promotion can be expected to eventually contribute to mitigating the overall GHG emissions in the mining sector if the GHG emission factor of China's power supply maintains the decreasing trend presented in the past decade.
Additionally, further suggestions aimed at mitigating GHG emissions in underground metal mines can be provided. Although existing literature has proposed numerous policies or technologies, such as taxing GHG emissions [53], carbon cap and trading [54], developing efficiency in diesel engines [55], and increasing the share of renewable energy in the power supply [16], the case studies suggest more elaborate mitigation approaches to these projects. For example, backfilling of the subsidence pit is suggested for cavingbased cases, so as to prevent the subsequent extension of surface subsidence after vertical caving. Such prevention has been validated in the Hemushan and Gongchanling iron mines [56]. Moreover, three approaches, i.e., using industrial wastes (such as fly ash or alkali-activated slag [57,58]) as cementing materials to reduce the consumption of cement in the backfilling stage, employing energy-efficient pumps, and further developing hauling systems, especially LHDs, are recommended for filling-based projects to mitigate their GHG emissions.
The uncertainty of this study should be clarified. Uncertainty primarily exists in the modelling for GHG emission estimation. The proposed model is a static input-output model, which means all input, e.g., emission factor of GHG sources and mining equipment, are assumed to be constant. However, these parameters are likely to varies along with the employment of GHG-abatement technologies [59] and the advance of power plants and oil refineries. The dynamic analysis, especially considering the potential of underground metal mines to mitigate GHG emissions, can be applied to deal with this uncertainty. Another uncertainty source is the variation of backfilling materials. The filling-based cases involved (i.e., Luohe and Longtangyan) use cemented materials, and plenty of electricity is consumed to produce, transport, and dewater the backfilling slurry. Besides the cemented materials, dry materials (e.g., waste rocks and tailings) is an alternative option for filling-based projects, although they are rarely used. This uncertainty can be significant, and detailed analysis is still required in the future, to provide comprehensive insight for policy decision.
Finally, the contributions of this study should be addressed. Theoretically, this study provides a unified model for obtaining an engineering-based inventory of the GHG emissions generated in underground metal mines by either the caving or filling method. Such an inventory is valid not only for designing macro-policy, but also for guiding mining projects to implement carbon-abatement technologies, or other actions for GHG emission mitigation or carbon intensity reduction.

Conclusions
The filling method is regarded as a cleaner approach for underground metal mines, and has been promoted in China. However, quantitative assessment for supporting such a promotion are rare. To conduct such an assessment, this study proposes a bottom-up model for estimating GHG emissions from caving-based or filling-based projects. The GHG emissions of the Luohe, Longtangyan, Maogong, and Xiaowanggou iron mines are estimated accordingly.
In contrast to popular belief, the results regarding GHG emissions from the case studies show that filling-based projects (17.386 and 15.211 t CO 2 eq/kt in Luohe and Longtangyan, respectively) generate more GHG emissions to produce unit mineral products than the caving-based projects (5.554 and 5.602 t CO 2 eq/kt in Maogong and Xiaowanggou, respectively) from a life-cycle perspective. Although the direct GHG emissions of filling-based projects are lower, the higher indirect values indicate that the emissions are transferred upstream to the cement and power sectors. This means that promoting the filling method is not a climate-beneficial decision, as such promotion will raise the life-cycle GHG emissions of China's domestically extracted mineral products.
Additionally, some approaches for GHG emission mitigation have been suggested, such as backfilling subsidence pits for caving-based projects, using industrial waste as cementing materials, and employing energy-efficient pumps for the filling-based.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to request by the involved cases.