Since the founding of New China in 1949, nationwide construction has spurred demand for coal and the exploitation of coal resources. Currently, coal production in China accounts for almost half of the world’s total production [1
]. Since 2005, to improve the efficient utilization and safe mining of coal resources, China has carried out the effective integration of coal resources [2
]. However, most coal mines with integrated resources exhibit problems, such as a serious lack of data on the old goaf, unclear distribution of the boundary area, unclear water area and quantity in the goaf, and difficulty carrying out geophysical prospecting work underground in small closed-pit coal mines, which have exposed serious hidden dangers to coal mine safety. The Pingshuo Dong open-cast coal mine is a typical example. Due to long-term disorderly mining, no-filling mining, rich mining, and poor abandonment of surrounding small coal mines, many underground goafs exist in the Pingshuo Dong open-cut coal mine, and most of these goafs have no recorded specific information, such as location and size. If the roof thickness of the goaf is less than the minimum safe thickness, a roof collapse can occur, which can cause surface subsidence or cracking, seriously threatening the safety of personnel and equipment working above the roof [3
Geophysical prospecting methods commonly used in goaf and water accumulation include the shallow seismic method [5
], transient electromagnetic method [7
], high-density electrical method [10
], magnetotelluric method [12
], and ground-penetrating radar [14
]. Various geophysical methods have achieved certain effects in detecting goaf and water accumulation, but each geophysical method has its own applicable conditions and advantages. The shallow seismic exploration method allows continuous measurement, and the range of goaf detected is more accurate and reliable, but it can hardly reflect the accumulated water of the goaf area [16
]. The transient electromagnetic method is especially sensitive to low resistance and has a good effect in detecting goafs with accumulated water, and the detection depth is relatively large, but it is greatly affected by electromagnetic interference. The high-density resistivity method has a better detection effect on goaf water accumulation, but it is affected by topography and grounding resistance [17
]. The magnetotelluric method is effective at detecting water accumulation in goafs, and the detection depth is large, but it can be affected by static effects and near-field effects. Because a single geophysical method is limited, the geophysical interpretation often exhibits multiple solutions and low reliability [18
]. In recent years, many studies and applications have been carried out for integrated geophysical methods [20
]. Integrated geophysical prospecting can obtain more physical information, verify and supplement other methods, reduce the multiple solutions of a single geophysical prospecting method, and improve the resolution of exploration and the reliability of interpretation results. However, for open-pit mining areas, the terrain conditions are more complicated, the coal seams are very shallow, and there are high-drop steps (transport lines). For goaf distribution and water accumulation prediction, the high-density electric method cannot meet the condition of the line layout, the ground-penetrating radar detection depth is insufficient, and the shallow seismic acquisition coverage is uneven. How to achieve higher resolution and better exploration results has not been systematically discussed in depth. This paper takes the prediction of the goaf distribution and water accumulation of the Pingshuo Dong open-cast coal mine as an example. Aiming at this type of problem, this paper proposes combining small-array high-density acquisition, multidomain joint denoising, and bin homogenization, weakening the influence of high-drop steps on shallow seismic exploration, and then using the multiple-attribute method to predict the goaf. Based on the electromagnetic environment analysis and topographical features of the study area, the transient electromagnetic method was used to identify goafs. The practical application shows that the high-density 3D seismic and transient electromagnetic comprehensive interpretation can effectively identify the goaf and water accumulation characteristics of an ultrashallow high-altitude open-pit mine. This proves the feasibility and application of the proposed method and can provide a meaningful reference for research in similar areas.
4. Application and Discussion
The high-density three-dimensional seismic and transient electromagnetic combined detection method is employed to identify the goafs in the detection area. Three-dimensional seismic design has 5843 survey points in the whole area, including 2399 shot points and 3444 detection points. The exploration area is 0.25 km2
, and the coverage area of 24 times is 0.2 km2
. The normal distance of the transient electromagnetic method is 26~35 m. The engineering survey line is arranged along the design roadway and its surroundings. The point distance of key areas above the roadway is 10 m, some areas are densified to 5 m, and the point distance of other areas is 20 m. The survey lines are arranged along the east–west axis, and a total of six survey lines are arranged. Two survey lines are arranged above the No. 4 coal roadway, numbered N0 and N35 from south to north. Four survey lines are arranged along the No. 9 coal roadway, numbered l0, L33, l59, and L85 from south to north, and the survey points are numbered 0, 20, and 40 from west to east according to the point distance, as shown in Figure 4
The buried depth of Coal Seam 4 is 0–150 m, with an average of 70 m. The 9th coal seam has a buried depth of 40–210 m, with an average of about 110 m. The extremely shallow burial depth of the target layer and the small survey area led to the limited length of survey lines. Therefore, seismic exploration adopts the method of small array and high-density acquisition, with high shot-point density and high coverage times. After the field collection test, the shot interval has changed from 24 m to 6 m, and the coverage times have changed from 24 to 64. Meanwhile, due to the restriction of the field acquisition and construction conditions (vehicle walking platform), the distribution of shot and detection points is extremely uneven (Figure 5
a), resulting in extremely uneven coverage of the study area (Figure 5
Due to the limitation of terrain conditions, such as large drop and steep slopes between the platforms, the original coverage times are extremely uneven. Therefore, the surface homogenization method is used to eliminate the adverse effects of uneven coverage, including static bin homogenization and dynamic bin homogenization. Static bin homogenization is used to make only one track in each offset group in a bin. If there are N tracks in an offset group in this bin, the N tracks in the offset are weighted and superimposed to represent the offset group; if there are N tracks in an offset group in the bin, then the N tracks are weighted and superimposed, and the superimposed track represents the offset group. Dynamic surface equalization is mainly used to find the offset group in the adjacent element. Define a large rectangular bin centered on the current bin. Any missing offset set in the current bin searches for the corresponding offset set in the adjacent bin within a circle, where the radius of the circle is a linear function of the offset set and its range is limited to the large bin.
During the processing, the seismic trace is edited on the single-shot record to eliminate the strong interference channel and the invalid redundant channel. On this basis, dynamic panel homogenization is used to supplement the missing offset set, and static panel homogenization is used to make the coverage times and energy more balanced (Figure 5
shows the CDP gathers before and after bin homogenization, where Figure 6
a,b show the processing effects of bin-homogenizing the number of insufficient superpositions traced to the average number of superpositions. Figure 6
c, d show the processing effects of bin-homogenizing the number of excessive superpositions traced to the average number of superpositions. From the processing effects, we find that the basic shape of the seismic profile has not changed. The overall energy of the seismic section is more balanced.
shows the comparison of stacked gathers before and after bin homogenization. We can determine that the energy of the missing trace is recovered (at the red elliptical mark), and the amplitude of the homogenized bin can reflect the real situation of the underground media.
Due to the influence of engineering construction, noise has a great influence on shallow seismic exploration. Suppressing noise is very important for the accurate identification of goafs. Data analysis shows that the main noise sources include linear interference, surface waves, and abnormally strong amplitudes. In this study, a multidomain joint denoising method is employed through experiments to classify and eliminate different types of noise. This follows the law of progressive, step-by-step denoising and gradually improves the signal-to-noise ratio. For linear interference, the apparent speed and frequency range of linear interference is analyzed, and the noise attenuation method with less waveform modification is adopted to eliminate the noise. For the outliers and abnormal amplitudes, the regional anomalous amplitude attenuation method and the frequency-division abnormal amplitude attenuation method are used for suppression. The strategy of multiple-trace statistics, single-trace denoising, and frequency-division suppression is adopted to suppress random abnormal amplitude noise and surface waves. Strong energy interference in seismic records is identified automatically in different frequency bands, and the spatial position of noise is determined. According to the defined threshold value and attenuation coefficient, time-varying and space-varying methods are adopted to suppress random abnormal amplitude noise. Figure 8
shows the comparison of single-shot records before and after suppressing noise. Within the scope of the red rectangular box, we can determine that the surface waves and other noises are effectively suppressed, and the energy of the primary reflection wave is obviously enhanced. Figure 9
is the comparison of stacked gathers before and after suppressing noise. Within the scope of the red elliptical frame, it can be seen that the weak waveform after noise suppression is displayed, which can more effectively reflect the subtle changes in underground media.
This study identifies goaf and water accumulation characteristics by three steps: seismic forward modeling assistant interpretation, goaf seismic-wave-anomaly characteristics interpretation, and seismic- and electrical-method comprehensive-anomaly interpretation. First, a goaf geological model suitable for the exploration area is established, and wave forward modeling is carried out to study the seismic response characteristics of various goafs (Figure 10
) and assist in anomaly interpretation.
Due to the air or water content in the goaf, the absorption coefficient of the goaf is significantly higher than that of the normal coal seam, resulting in no reflection or weak reflection at the bottom interface of the goaf and substratum [32
]. There is a fault in the wave group; the reflected wave is characterized by energy enhancement, frequency reduction, and an obvious concave phase (Figure 11
a), or the energy becomes weaker and the continuity becomes worse (Figure 11
Comparing the reflected waves, seismic attributes, and drilling verification results of coal seams in the study area, a total of four coal-seam goafs are explained. The goaf labeled ① is located at the northern boundary of the exploration area, and the goaf anomaly is irregular. Combined with the interpretation results of the northern transient electromagnetic method, the goaf is produced by mining from north to south in the Yuling coal mine, and the abnormal area in the seismic work area is approximately 1359 m2.
The goaf marked with ② is located outside the southeast exploration boundary, in front of the Xiaoxiyao mine roadway, and the boundary of the abnormal area is irregular. The goaf is caused by mining from east to west in the Xiaoxiyao mine, and the abnormal area in the area is approximately 7352 m2.
The goaf marked with ③ and ④ is located southeast of the exploration boundary, and the abnormal boundary is irregular. The goaf is formed by the westward mining of the Xiaoxiyao mine. However, the northern part of the abnormal area is greatly disturbed by the raw coal transportation channel, and the southern part may be affected by the deterioration of data quality due to its close proximity to the boundary of the detection area, so the reliability of the results needs to be verified by more drilling results.
There is a seismic anomaly zone of coal-seam structure change in coal seam 9, which is located in the west of the exploration area. The plane shape is an irregular polygon, and the east side is connected with the seismic anomaly of the collapse column. The scope is delineated by a dotted line in Figure 12
. The drilling data confirmed that the reason was that coal seams 9–1 and 9–2 merged into one layer.
A collapse column was identified via the drilling and log data from coal seam 4 (Figure 13
). The lithology below K2 sandstone, as the marker bed of the No. 4 coal roof, is breccia, accompanied by a small amount of coal chips and blocks, with poor cementation. The shape of the collapse column on the plane is an approximate ellipse, especially in the instantaneous frequency attribute, and the characteristics of the collapse column are the most obvious. The collapsed column is a good water channel and water-accumulating area, so it is very important to identify the development of the coal-seam collapse column.
Based on the seismic interpretation of the development characteristics of special geological bodies such as goafs, faults, and collapse columns, the transient electromagnetic method is employed to predict the water accumulation in the key areas of the study area. The transient electromagnetic survey lines covering this area are N0 and N35. Due to terrain limitations, we adopted small coils and multiple circles of central loop to detect water accumulation in the goaf using the sensitivity of electrical resistivity to water.
The length of survey line N35 is 910 m, and it is located on the 1320 platform. There are two small-scale low-resistance anomalies with relatively small ranges and amplitudes, which are located near point 310 and point 570 (at the red circle of the Figure 14
). Among them, the low-resistivity anomaly of measuring point 570 is also shown near the corresponding position of line N0, and the anomaly is caused by the enhanced water yield of the formation near the fault. The abnormal position of point 310 is consistent with the position of the collapse column interpreted by the seismic attribute, which is speculated to be caused by the increased water abundances of the collapse column.
The length of survey line N0 is 790 m, which is located on the 1350 platform. The apparent resistivity profile east of survey point 530 (in the direction of the large point) shows a large low-resistivity anomaly (at the red circle of Figure 15
). Combined with the comprehensive interpretation of adjacent survey line N35 and three-dimensional seismic data, it is speculated that the anomaly here is caused by the fragmentation of the formation near the fault and the enhancement in water abundances.
Integrating the results of the seismic interpretation of the goaf and the transient electromagnetic interpretation of the low-resistance abnormal area of each survey line, the scope of the goaf and water accumulation, collapse column, and fault water-rich anomaly displayed in the detection area is delineated on the plane. The subsequent drilling and testing results are basically consistent with the prediction results; in particular, the accurate prediction of the scope of the goaf plays a vital role in safe mining and personnel safety. Figure 16
shows the verification hole DLT-18-6 passing through goaf No. 4. The verification hole DLT-18-6 in the goaf with number 4, compared with the information of the adjacent well, revealed that the section where no coal core was found happened to be the depth section of coal 4 and 9. Due to the presence of goaf, the diameter value of the coal-seam section shows significant fluctuations, and the gamma–gamma value is significantly lower than that of the No.3 coal seam. Combined with the prediction of water accumulation, relevant coal-mining measures can be effectively formulated to avoid production risks (Figure 17
The local part of the study area is particularly affected by the on-site construction conditions, such as the mining party’s heavy vehicles, gas stations, drilling areas, and blasting areas, which are close to the geophysical exploration work area or even overlap and have a large interference with the seismic signal. In addition, the local coverage times are low, and the signal-to-noise ratio of the data is poor, which directly leads to the poor reliability of the interpretation results. For example, the southwest corner of the study area is a typical disturbed area, with a large area of abnormal areas. However, we predicted the possible results in advance, and the results were confirmed by a very small amount of drilling. Therefore, the judgment of the reliability of data is a very important link.
To ensure safe and efficient mining, it is recommended to deploy ground drilling or advance drilling and geophysical exploration in the abnormal area and poorly controlled area in advance during the mining process to further determine the nature and location of the abnormality and ensure safety [33
]. Combined with relevant geological data such as tunneling and drilling in different stages, comprehensive geological research should be strengthened to further improve the reliability and accuracy of interpretation results.