Characteristics of Overburden Damage and Rainfall-Induced Disaster Mechanisms in Shallowly Buried Coal Seam Mining: A Case Study in a Gully Region

Abstract

Shallow coal mining in gully regions has resulted in significant subsidence hazards and increased the risk of surface water inflow into mining panels, compromising the sustainability of surface water management and underground resource exploitation. In this study, the chain disaster process caused by shallow coal seam mining and heavy rainfall is quantitatively analyzed. The findings reveal that shallow coal seam mining leads to the formation of caved and fractured zones in the vertical direction of the overlying rock. The fractured zone can be further classified into a compression subsidence zone and a shear subsidence zone in the horizontal direction. The shear subsidence zone is responsible for generating compression and shear deformations, intercepting rainfall runoff, and potentially triggering landslides, necessitating crack landfill treatments, which are critical for promoting sustainable mining practices. The HEC-RAS program was utilized to integrate annual maximum daily rainfall data across different frequencies, enabling the establishment of a dynamic risk assessment model for barrier lakes. Numerical simulations based on unsaturated seepage theory provide insights into the infiltration and seepage behavior of rainfall in the study area, indicating a significant increase in saturation within lower gully terrain. Precipitation infiltration was found to enhance the saturation of the shallow rock mass, reducing matric suction in unsaturated areas. Finally, the disaster chain is discussed, and recommendations for managing different stages of risk are proposed. This study offers a valuable reference for the prevention and control of surface water damage under coal mining conditions in gully regions.

1. Introduction

Mine water hazards are one of the primary disasters limiting sustainable resource extraction. Water inundation incidents in Chinese coal mines arise from various water-related threats, with over 30 distinct types, including surface water, pore water, and karst water, among others [1]. Surface water damage, accounting for 10% of total water-related accidents, is a significant issue in mining regions with shallow coal seams and surface gullies, particularly in northwestern and northern China [2,3]. During the rainy season, water inrush from roof strata following the mining of shallow coal seams (less than 150 m deep) in gully regions is a prominent concern in northwestern and northern China. As shown in Figure 1, the goaf formed after mining the upper coal seam creates space for water accumulation, exacerbated by surface runoff during floods. Due to the proximity of upper and lower coal seams, fractured roofs exhibit limited structural integrity. This weakens the boundary between the upper seam’s damaged floor and the lower seam’s water-conducting fractured zone, significantly increasing permeability in the coal and rock strata [4,5], resulting in disasters. Three key factors contribute to the regional geological and ecological challenges: (1) shallow coal seam mining creates pathways for surface water infiltration; (2) gully topography increases rock movement complexity; and (3) abundant rainfall during the flood season serves as the primary water source. Subsidence of rock strata caused by coal extraction is the fundamental cause of mining disasters, which facilitates the formation of hazardous conditions within mines [6,7,8,9]. Therefore, studying the overburden failure characteristics and rainfall-induced disaster mechanisms in shallow coal seam mining within gully regions is crucial for ensuring both resource and environmental sustainability.

Research on mining-induced hazards has primarily focused on individual disasters, yielding significant insights into the mechanisms and movement processes of single hazards. In terms of mine water damage, research primarily focuses on predicting water-conducting fractured zones and mapping overburden strata failure [10,11,12], which are delineated based on the deformation, fracturing, and hydraulic conductivity characteristics of the overlying rock [13,14]. The gully topography engenders intricate rock kinematics and surface deformations resulting from coal mining operations [15,16]. However, most research has focused on the partitioning characteristics of overburden in the vertical direction, with comparatively little attention given to the horizontal partitioning characteristics of overburden caused by mining activities in gully areas. Understanding the horizontal deformation and failure characteristics of gully terrains is essential for developing targeted disaster mitigation strategies.

Current research on strata movement induced by shallow coal seam mining in gully terrains largely centers on fracture propagation in the overlying rock strata and surface [17,18,19,20], polygon block hinged structures of slopes [21,22,23], stability analysis of the overlying slope in the goaf region [24], activation of landslide by slow subsidence [25,26], and nonconventional subsidence effects, such as valley closure [27,28]. Furthermore, it provides favorable geological circumstances for flood accumulation and the occurrence of barrier lake disasters during the flood season [19,29,30]. This confluence of factors—including rainfall, gully topography, and shallow coal seam mining—can lead to cascading disasters such as crack propagation, landslides, and intensified water bursts that block river channels. Rainfall infiltration in slope engineering can be effectively simulated using unsaturated seepage theory [31]. Most existing research focuses on two-dimensional slope stability and the interactions between seepage, stress, and damage [32,33,34]. Wu and Selvadurai [35] developed a two-dimensional coupled flow-deformation model to describe groundwater level variations induced by rainfall, examining the impact of boundary conditions on the coupled equations of unsaturated flow and deformation, based on in situ data from a slope in South Korea and real-time precipitation measurements. Similarly, Kim and Jeong [36] utilized a coupled model to investigate the hydromechanical response of shallow landslides in unsaturated slopes during rainfall infiltration, detailing the failure process of slopes under transient conditions. Limited attention has been paid to unsaturated rainfall infiltration in three-dimensional space under mining conditions. Despite recognition of the complexity of gully disaster chains, comprehensive monitoring of these disaster chains remains challenging, especially with the sudden and unpredictable formation of barrier lakes in gully areas. The use of Geographic Information System (GIS) technology and numerical simulation helps address these limitations. In recent years, numerical simulations have become the dominant approach for studying geological hazard mechanisms and conducting quantitative assessments. Discrete element methods, such as UDEC and 3DEC, effectively capture the progressive failure processes in mining, including crack propagation, rock collapse, and sliding, providing deeper insights into rock deformation and stress distribution [37,38,39]. GIS and hydrodynamic models have been widely adopted for flood studies in mining areas, producing promising results [40,41,42]. The Hydrologic Engineering Center’s River Analysis System (HEC-RAS) is a widely recognized and extensively used model for simulating small watershed floods [43,44]. Finite element methods, like COMSOL and FEFLOW, allow for the creation of detailed three-dimensional geological models to simulate groundwater flow and its response to rainfall, aiding in the prediction of potential water-related hazards [45,46].

The mining-induced subsidence in gully areas can trigger a cascade of secondary disasters, including landslides, landslide-dammed lakes, and runoff infiltration. These disaster chains form rapidly, with sudden onset, and pose catastrophic risks to underground coal mines. Despite a comprehensive literature review, we found that there is still a lack of analysis on the entire process of disaster chains induced by surface subsidence and rainfall in coal-mining areas. To address this gap, our study employs the discrete element method to analyze the characteristics of overburden failure caused by mining in gully terrains. The vertical zoning features serve as the foundation for numerical simulation of rainfall infiltration, while the horizontal zoning characteristics aim to guide the implementation of surface water seepage prevention measures in mines. Actual rainfall data is used to simulate the rainfall-runoff process in the study area, providing mine engineers with a clear assessment of the risks associated with surface water accumulation above the mining panels. Additionally, the unsaturated rainfall infiltration process is investigated using the finite element method. This case study aims to provide a reference for preventing and controlling runoff disasters caused by coal extraction in gully areas, promoting the sustainable management of coal mining and surface water resources.

2. Engineering Background

The research area is located in Panel No. 28306 of the Xiqu Coal Mine in Shanxi Province, China. The mine contains seven distinct geological units, each characterized by unique structural and geomechanical properties. These lithological formations exhibit varying resistance to weathering and erosion. The mining area features a complex gully terrain, with significant surface water accumulation during the flood season (Figure 2). The mined coal seam belongs to the Carboniferous-Permian system. Most of the resources in the upper Nos. 2 + 3 coal seams at Xiqu Coal Mine were extracted before 2010, with current mining focused on the lower No. 8 coal seam.

Before mining Panel No. 28306, the extent of the goaf and water accumulation area in the Nos. 2 + 3 coal seam was delineated through underground transient electro-magnetic detection and surface surveys. Five anomalous water accumulation zones were identified within the goaf of the Nos. 2 + 3 coal seams, and active drainage measures were implemented. Water samples were collected from drilling holes for subsequent quality analysis. The water quality was classified as SO₄ · HCO₃-Ca-Mg type with a pH range of 6.5 to 7.2. In a specific water exploration hole, upon initial opening of the hole, the mineralization level reached 3379 mg/L. During the water discharge process, the mineralization level significantly decreased and stabilized around 1000 mg/L. As the accumulated water approached drainage, the mineralization level increased notably, reaching 1453 mg/L. The analysis suggests that upon the initial opening of the water exploration hole, the stagnant accumulated water in the small kiln empty roadway caused the sedimentation of various ions, resulting in higher mineralization levels during exploration. During the water discharge process, the water output from the exploration hole remained stable, leading to a decrease in ion content and subsequent mineralization decrease and stabilization. As goaf water was drained, there was a significant change in water quality, leading to increased solubility of ions and a substantial increase in mineralization. During this period, the relative content of Na+ increased, while the relative levels of SO₂- and HCO₃- decreased, indicating a transition toward surface water characteristics. During inspections conducted by technical personnel, it was observed that the drainage hole’s water volume peaked during exploration and gradually decreased. After three to four days, the water volume was essentially negligible. However, due to continuous rainfall in the area, a small amount of water was discharged from each drainage hole at rates ranging from 0.5 to 2.0 m³/h. This could be attributed to delayed water supply from the abandoned Nos. 2 + 3 coal seam tunnels caused by rainfall. A comprehensive analysis of water quantity and quality changes across the boreholes confirmed a dynamic water supply from the abandoned roadway above Panel No. 28306.

Panel No. 28306 was completed in 2021, with mining conducted from the northern to southern direction. The overburden above the panel is approximately 90 to 188 m thick. To the west is the goaf of Panel No. 28307, which was mined out in 2017, while the unmined Panel No. 28305 lies to the east. The Nos. 2 + 3 coal seams near the gully area were mined using the residual pillar method by small coal mines. The No. 8 coal seam is 4.10 m thick and dips southwest at an average angle of 4°. The inclined longwall retreat method, with one-time full-height mining and full-collapse mechanized mining, is employed.

As shown in Figure 3, there are five relatively thick roof layers (each less than 10 m thick) within a 143-m range above the No. 8 coal seam. These include three layers of sandy mudstone, which are dense and lack stratification, contain small amounts of pyrite grains, and have occasional incomplete plant fossils. The upper portions of these layers are loose and friable. Additionally, there is a 7.55-meter-thick mudstone layer, characterized by its fine, hard texture with slight sandy content and indistinct stratification. Finally, there is a 5-meter-thick fine sandstone layer, primarily composed of quartz with thin argillaceous cementation. The 19.64-meter-thick sandstone roof is dark grey to grey-black, primarily composed of quartz. It is relatively hard, dense, and has few pores or cracks, qualifying it as a key stratum (KS) between coal seams [47,48]. The overall roof strata of the No. 8 coal seam consist of hard rocks, while the overlying layers of the Nos. 2 + 3 coal seams are composed primarily of alternating sandstone and mudstone.

3. Materials and Methods

The current study conducted an on-site investigation of geological hazards induced by underground mining in the study area. Crack and fissure propagation in the gully terrain, resulting from mining activities, was analyzed using 3DEC 5.2 version software. Additionally, a physical simulation experiment was performed to examine the fracture zoning characteristics of overlying strata in the gully area. The rainfall-runoff process was simulated using HEC-RAS, based on actual rainfall data, to assess the risk of surface water accumulation above the mining panel. The unsaturated module of COMSOL Multiphysics was then employed to analyze the characteristics and mechanisms of rainwater infiltration in the mining area.

3.1. Site Investigation and Distinct Element Method

Figure 4a presents the constructed numerical model, which uses the I-I’ profile from Figure 2. The distance between the ground surface above the east roadway of Panel No. 28306 and the nearest point of the Fanshigou gully channel is only 90 m, with several planned mining areas, including Panel No. 28305, located directly beneath the Fanshigou gully channel (Figure 4a). The mining of Panel No. 28306 triggered a landslide, resulting in road blockages due to large fallen rocks (Figure 4b) and surface collapse (Figure 4c). Additionally, during the flood season, the Fanshigou gully channel, located east of Panel No. 28306, was completely submerged, with accumulated water depths ranging from 3.5 to 7.2 m (Figure 4d). Fractured rock masses, commonly found in natural environments, require careful integration into numerical simulations due to their discrete characteristics [49]. The three-dimensional distinct element code (3DEC) is widely used to analyze the progressive failure of rock strata, providing enhanced capabilities in replicating discontinuous material features throughout the failure process [28,50]. To investigate the progressive failure of the overburden during the mining of Panel No. 28306 at Xiqu Coal Mine, a numerical model (Figure 4a) was constructed based on the elevation contours and borehole data from the I-I’ profile (Figure 2). The 3DEC simulation accounted for the working conditions of residual pillar mining in the upper Nos. 2 + 3 coal seams.

The numerical model includes five primary rock types: sandstone, sand-mudstone interbeds, sandy mudstone, coal, and limestone. The mechanical parameters of these rock types and their joints are provided in

Table 1. Mechanical parameters applied in numerical calculations.

Lithology	Bulk Modulus (GPa)	Shear Modulus (GPa)	Density (kg·m−3)	Cohesion (MPa)	Tensile Strength (MPa)	Internal Friction Angle (°)
Sandstone	16.7	7.7	2700	5.6	3.5	29
Sand-mudstone interbed	13.4	6.62	2600	4.95	3.25	31
Sandy mudstone	14.2	6.5	2520	4.36	3.2	34
Coal	5.2	2.78	1460	2.4	1.28	25
Limestone	16.57	4.53	2750	5.9	3.9	38

and

Table 2. Mechanical parameters of the joints.

Joint Material	Joint Normal Stiffness (GPa)	Joint Shear Stiffness (GPa)	Joint Cohesion (MPa)	Joint Friction Angle (°)	Joint Tensile Strength (MPa)	Joint Type
1	140	14	2	28	1	Bedding joint
2	100	10	4	35	2	Vertical joint

, with validation supported by prior studies of the mine [48]. The Mohr-Coulomb model was applied to the block model, while the joint model was based on the Coulomb sliding model. The model’s boundaries are constrained, with normal displacement limited on all sides, and horizontal and vertical displacement restricted at the bottom, while the ground surface remains a free boundary. Mining proceeds from the left to the right side of the coal seam.

3.2. Similar Simulation Experiment

To investigate the overburden zoning characteristics in coal seam mining under gully terrain, a simulation test on similar materials was conducted, based on the II-II’ section depicted in Figure 2, which represents the actual mining of Panel no. 28306. Quartz sand was employed as the aggregate, while gypsum and lime served as binding agents. Additionally, mica powder was inserted between the strata. The physical model had dimensions of 2.5 m in length, 1.0 m in height, and 0.15 m in thickness. By applying a geometric similarity ratio of 1:150, the simulated dimensions of the actual strata were scaled to 375 m in length and 150 m in height. The actual rock mass exhibited an average density of 2.4 g/cm³, while the simulated rock mass possessed an average density of 1.65 g/cm³, resulting in a density similarity ratio of K = 1.45:1.

3.3. Rainfall Runoff and Infiltration Simulation Method

Understanding the characteristics of surface catchment flooding in gullies during rainstorms is a key prerequisite for preventing surface water disasters. A rainfall-runoff simulation for the study area was performed using HEC-RAS 6.2 version, an open-source software compatible with GIS that supports one-dimensional steady-state hydraulic calculations, rainfall-runoff simulations, and one- and two-dimensional unsteady flow calculations [42,51,52]. High-precision terrain data were obtained using oblique photogrammetry with an unmanned aerial vehicle (UAV), ensuring accurate simulation results for rainfall-induced flooding. Over 3000 photographs were taken within the study area by UAV. The drone maintained a constant altitude of 100 m and a speed of 5 m per second throughout the flight. The high-resolution terrain model used for the rainfall-runoff simulation, in GeoTIFF format, was generated using DJI Terra 3.0.0 version software. As shown in Figure 5, the spatial step size of the 2D grid is set to 4 m, and a total of 50,328 grids are generated in the study area. The downstream boundary of the gully’s riverbed slope (Normal Depth) was set to 0.0015 based on the terrain conditions. Surface rainfall was defined as the inflow boundary condition to simulate the impact of watershed precipitation on the hydrological process.

Figure 6 shows the COMSOL numerical model and grid used to simulate rainfall infiltration. COMSOL’s capability to model the nonlinear behavior of unsaturated flow, combined with its flexibility in handling varying boundary conditions and material properties, makes it ideal for predicting infiltration patterns and assessing the risk of rock and soil instability [32,35]. The model dimensions are 1200 m (x-direction), 500 m (y-direction), and 204 m (z-direction). The vertical overburden was divided into distinct zones: the repeated disturbance zone, fractured zone, caved zone, coal seam No. 8, floor failure zone, and bedrock zone, based on overburden fracturing characteristics following coal seam mining (Section 4.1). Permeability coefficients and porosity values for each rock stratum were obtained from drilling and pumping tests, as summarized in

Table 3. Permeability coefficient and porosity values of rock strata.

Thickness (m)	Lithology	Strata Classification after Coal Mining	Hydraulic Conductivity (m/s)	Porosity
70	Sand-mudstone interbed	Repeated disturbance zone	1.704 × 10−5	0.24
9	Sandy mudstone
17	Sand-mudstone interbed
4	Nos. 2 + 3 coal seam
15	Sand-mudstone interbed	Fractured zone	2.336 × 10−5	0.22
3	No.4 coal seam
5	Sandy mudstone
9	Sand-mudstone interbed
20	Sandstone
2	Limestone	Caved zone	5.247 × 10−5	0.32
7	Sandstone
3	Limestone
7	Sandstone
4	Limestone
4	No. 8 coal seam	No. 8 coal seam	7.018 × 10−7	0.15
12	Sand-mudstone interbed	Floor failure zone	2.036 × 10−5	0.20
2	No. 9 coal seam	Bedrock zone	4.522 × 10−7	0.12
11	Sand-mudstone interbed

. In the rainfall infiltration simulation, it was assumed that no aquifers exist above the No. 8 coal seam roof, and the surrounding areas lacked recharge boundaries. The bottom surface of the simulation area was treated as an impermeable boundary, while the top surface served as the replenishment boundary, receiving rainfall input. To enhance computational convergence, uniform rainfall with a constant intensity of 3.67 mm/h—equivalent to the rainfall recorded on 4 October 2021—was applied over a total duration of 96 h. The following assumptions were considered in the numerical simulation: (i) isotropic spatial permeability coefficient of the same rock stratum; (ii) incompressible fluid adhering to Darcy’s law; (iii) rainfall does not affect the geometric shape of the three-dimensional gully; (iv) neglect of slope failure induced by rainfall, with only linear elastic parameters of solid mechanics being set; (v) all rainfall infiltrates.

4. Results

4.1. Characteristics of Surface Subsidence and Crack Development

To investigate the overburden zoning characteristics of coal seam mining in gully terrain, a residual pillar mining method was applied to the upper Nos. 2 + 3 coal seams. A 20-m coal pillar was left after every 20 m of mining. This approach significantly reduced stress concentration and minimized the collapse of the overlying rock (Figure 7). Field surveys revealed that, after the extraction of the Nos. 2 + 3 seams, no obvious stepped cracks or collapse pits were observed. However, underground mining activities weakened the shear strength of the shallow slope layers, leading to the formation of tensile cracks, which created preferential pathways for rainwater infiltration. As a result, landslides occasionally occurred due to the combined effects of mining and rainfall. Subsequent mining of Panel No. 28306, which advanced towards the Fanshigou gully, activated the shallow Nos. 2 + 3 coal seam goaf due to repeated mining and key strata (KS) failure. This resulted in an increase in stepped cracks, an expansion of collapse pits, intensified surface subsidence, and the formation of numerous cracks on the gully slope, which became the main infiltration channels for surface water. Due to the shallow burial of the coal seam, the overlying rock did not exhibit the typical “three zones” (caved, fractured, and continuous deformation zones). Instead, only two zones were observed: the caved zone and the fractured zone. The fractured zone extended to the surface, connecting with the Nos. 2 + 3 goaf. The caved zone of the No. 8 coal seam reached a height of 23 m, concentrated below the KS (Figure 7).

Table 4. Overview of each mining panel.

Coal Mine	Overview of Mining Panels							Height of Caved Zone/m
	Mining Panel	Length/m	Mining Thickness/m	Burial Depth/m	Coal Seam Inclination Angle/°	Coal Seam	The Roof of the Coal Seam
Dongqu	No. 28806	226	4.83	333	4	No. 8	Limestone	20.98
	No. 28206	170	4.79	255	4	No. 8	Limestone	18.32
	No. 18311	179	3.77	231	4	No. 8	Mudstone	19.28
Malan	No. 10702	216	2.08	402	4	Nos. 2 + 3	Sandy mudstone	21.00
	No. 12511	196	2.47	359	3	Nos. 2 + 3	Sandy mudstone	22.00
	No. 18301	225	4.03	470	4	No. 8	Limestone	19.00
Tunlan	No. 28120	235	3.33	431	4	No. 8	Limestone	27.58
Guandi	No. 22611	220	2.99	679	6	Nos. 2 + 3	Sandy mudstone	20.70
Xiqu *	No. 28306	220	4.00	147	4	No. 8	Sandy mudstone, limestone	23.00

* The mining area in this paper.

summarizes the mining panels, showing that the height of the caved zone typically concentrated around 20 m. Subsidence induced by the extraction of coal seam No. 8 exhibited a peak range of 3.5 to 3.9 m, with a subsidence coefficient between 0.87 and 0.97, consistent with previously documented values for the region.

Previous research has addressed the phenomenon in shallow coal seam mining, where KS fractures occur when the span reaches a critical threshold [53]. This fracture causes subsidence of the rock mass between the KS and the surface. As illustrated in Figure 8a (highlighted in red), the rock mass beyond the arch shoulder becomes unstable and prone to sliding along potential failure planes. Figure 8c,d show the distribution of fractures induced by mining, with significant fragmentation of the lower KS and the development of surface cracks on slopes adjacent to the gullies due to mining-induced subsidence. Mining-induced deformation in gully terrain generally forms distinct zoning characteristics in the horizontal direction. These deformations in an “inverted trapezoidal” gully can be classified into three zones: shear subsidence, compression subsidence, and caved zones (Figure 8b). A “shear subsidence zone” is formed on both sides of the goaf, predominantly subjected to horizontal tensile stress and vertical compressive stress. This results in the overlying rock mass tilting toward the center of the goaf. The movement of the rock mass within the shear subsidence zone increases the lateral displacement toward the exposed face, generating tensile stress on the slope. Consequently, tensile cracks develop along the rear of the slope, leading to the formation of a shear failure zone on the slope surface. A “compression subsidence zone” is located above the collapse-fractured zone, showing the highest degree of surface subsidence with minor compression deformation. Its stress is mainly vertical tensile stress and lateral compressive stress.

The deformation and failure characteristics of the overlying rock strata after mining Panels No. 28305 and 28306 are illustrated in Figure 9. The fractures in both mining areas predominantly exhibit tensile characteristics, with vertical fractures concentrated within 23 m above the roof. Above the KS, the number of shear fractures increases, and fractures in the overburden continue to develop after KS rupture. Gravitational forces exacerbate shear displacement along the slopes. Surface cracks on the slopes connect with upward fractures in the overburden, forming water channels that intercept surface runoff. Panel No. 28305 shows distinct shear and compression subsidence zones, influenced by factors such as the distance between the coal seam and the KS, the thickness and strength of the KS, and the gully slope. Sensitivity analysis of these factors represents a valuable area for future research.

4.2. Rainfall-Runoff Inundation of Study Area

4.2.1. Characteristics of the Fanshigou Basin

This section analyzes the local watershed characteristics using high-precision DEM and ArcGIS 10.6 version software. The Fanshigou watershed experiences its highest rainfall from July to October, with an average annual precipitation of 511.5 mm. Precipitation-induced runoff ultimately flows into the Fenhe River, as shown in Figure 10a. Based on the characteristics of the Xiqu Coal Mine watershed, the entire watershed was subdivided, identifying the Fanshigou gully catchment area where Panel No. 28306 is located, covering approximately 5.44 km² (Figure 10b). Land use in this area is predominantly forests, grasslands, and arable lands, accounting for 43.47%, 22.23%, and 19.02%, respectively (Figure 10c). Different land use types directly affect surface hydrological processes, thereby influencing the generation and flow of runoff. Land use types impact surface roughness (e.g., grasslands, forests, farmland, urban areas), which is a critical parameter in the HEC-RAS model for determining water flow speed and runoff behavior. The roughness values for each land use type are provided in

Table 5. Land use status of Fanshigou Basin [54].

Type	Area (m2)	Proportion (%)	Manning’s Roughness Coefficient
Arable land	1,034,953.85	19.02	0.03
Woodland	2,365,485.31	43.47	0.15
Grassland	1,209,895.56	22.23	0.04
Shrubland	208,735.47	3.84	0.07
Wetland	785.90	0.01	0.05
Water	550.13	0.01	0.025
Artificial surface	78,432.98	1.44	0.035
Bare land	543,293.80	9.98	0.025

.

4.2.2. Rainfall-Runoff Simulation Based on HEC-RAS

A rainfall-runoff simulation for the study region was performed using HEC-RAS, which offers a unique flood visualization feature, including X-Y plots, cross-sections, profile plots, rating curves, inundation maps, and hydrographs of river systems. The inundation mapping, done in HEC-RAS Mapper, allows for animation and the addition of multiple background layers (terrain, aerial imagery, etc.). For this case study, global precipitation was set as the inflow boundary, and the river outlet was designated as the lower boundary. Additionally, the presence of a barrier lake, formed by a landslide, was included. Precipitation data from 3 October 2021 (00:00) to 6 October 2021 (23:00) were used for the simulation (Figure 11).

4.2.3. Analysis of Rainfall Flooding Results

Results from the unsteady flow simulations in a two-dimensional framework reveal the inundation range as of 6:00 a.m. on October 6 (Figure 12a), showing significant submergence of the river. Water accumulation in the barrier lake ranged from 4.0 to 8.6 m in depth, covering a large area (Figure 12b). The simulation also identified multiple localized water accumulations in tributaries. Figure 12c presents the simulated depth and coverage of inundation from runoff without landslide-induced blockage, highlighting the different impacts. Due to road obstructions caused by the rainstorm, model calibration options were limited. Therefore, the depth of water accumulation at the barrier lake was used as the evaluation benchmark. However, the absence of accurate water stage measurements during the storm hindered optimal model results using the HEC-RAS 2D model.

As the river’s flood discharge was obstructed, the water level continued to rise, leading to overflow that submerged factory buildings (Figure 13a). The small watershed is characterized by a sinuous river channel, rugged riverbed, steep longitudinal slope, pronounced river incision, and rapid runoff convergence. Analysis of flood flow velocity (Figure 13b) shows that the maximum upstream velocity in the barrier lake channel reached 0.17 m/s. Higher velocities were observed in the center of the channel compared to the riverbanks. Narrower sections of the river channel experienced the highest velocities, leading to rapid flow increases, elevated energy levels, and significant impacts on riverbanks, all of which contributed to landslide occurrences.

4.3. Rainfall Infiltration Characteristics of Mining Areas

4.3.1. Control Equation for Unsaturated Seepage in Porous Media

The mining of Panel No. 28306 poses a risk to the stability of the surface riverbed and may lead to gully water infiltration. Therefore, understanding the mechanism of surface water infiltration into the mining site during the rainy season is crucial. This section focuses on the seepage characteristics of unsaturated rocks under rainfall conditions, utilizing Rhino 7.0 version modeling software and COMSOL 6.2 version to develop a three-dimensional gully computational model. The study integrates the coupling of the seepage and stress fields, conducting both steady-state and transient analyses of the gully during rainfall.

Under rainwater infiltration, water movement within sloping rock follows a classic saturated-unsaturated seepage process, governed by the Richards equation. This equation, fundamental to unsaturated seepage theory, is expanded into a saturated-unsaturated seepage equation using pore water pressure (p) as the independent variable. In COMSOL, the Richards equation is applied in the form of Darcy’s law, serving as the control equation for unsaturated seepage.

$$\rho \left[{\displaystyle \frac{C_m}{\rho g}}+S_{\mathrm{e}}S\right]\frac{\partial p}{\partial t}+\nabla •\rho \left[-\frac{{к}_s}{\mathsf{\mu }}{k}_{\mathrm{r}}\left(\nabla p+\rho g\nabla D\right)\right]=Q_m$$

(1)

where ρ is the fluid density, g is the acceleration of gravity, C_m represents the specific moisture capacity, S_e denotes the effective saturation, S is the storage coefficient, p is the pressure, к_s gives the saturated permeability, k_r denotes the relative permeability, μ is the fluid dynamic viscosity, D represents the elevation, and Q_m is the fluid source (positive) or sink (negative).

4.3.2. Geometric Models and Parameters

To achieve coupling between Darcy’s law and solid mechanics, two interfaces are established in the simulation. The Darcy’s law interface handles unsaturated seepage calculations, while the solid mechanics interface computes the stress field. These two are coupled through custom parameter settings, forming three interface pairs. The initial seepage and stress fields are obtained using initial parameters, with the results from the first pair of interfaces used as inputs for the second. Rainfall boundary conditions are incorporated only in the third pair of interfaces, which analyzes the transient response of the overlying rock during rainfall and calculates the seepage field throughout the process.

The infiltration model for this rainfall event is large in scale, requiring consideration of matric suction effects. When matric suction is significant, effective stress can be high, and the presence of large negative surface pressures may affect model convergence. Given the unsaturated nature of the model, the initial pressure field is assigned a negative pressure, and boundary conditions reflect this by applying negative pressure heads. In COMSOL, the van Genuchten (VG) equation relates saturation (Se) to pressure head (H_p). The equation is expressed as follows:

$$S\mathrm{e}=\{\begin{array}{cc}{\displaystyle \frac{1}{{\left[1+{\left|\alpha H_{\mathrm{P}}\right|}^n\right]}^m}}& H_{\mathrm{p}}<0\\ 1& H_{\mathrm{p}}\ge 0\end{array}$$

(2)

where α is assigned a value of 0.91, n is set to 2, and m is designated as 0.5.

The mining enterprise provided an initial Se value of 0.37 before rainfall, with a corresponding pressure head of 3 m, resulting in a negative pressure head of 3 m applied to the model boundary.

4.3.3. Simulation Results of Rainfall Infiltration

Figure 14a–d illustrate the evolution of saturation profiles during continuous rainfall. Initially, saturation at the slope surface increases, followed by a rise in the unsaturated zone as the saturation line descends. Based on the initial pore pressure and the condition that the water pressure at the water surface is 0, a negative pressure zone is anticipated above the groundwater level. In unsaturated rock strata, the pore spaces contain not only water but also gas, and the interface between water and gas exhibits surface tension. As a result, the pore air pressure does not equate to the pore water pressure, and this disparity in pressure is referred to as the matric suction. Matric suction within the slope gradually decreases. After 24 h of rainfall, saturation significantly increases in low-lying gully areas. By 48 h, saturation at the top of each profile rises notably, while the saturation line at the bottom of the slope drops. After 96 h of rainfall, lower-lying gully terrains exhibit a marked increase in saturation due to the higher hydraulic conductivity in the fractured strata caused by mining. Consequently, surface water does not continue to rise but instead infiltrates and spreads towards lower areas, with maximum saturation levels increasing from 0.34 to 1.0. Gully terrains, characterized by steep gradients and concave shapes, facilitate water accumulation and flow, amplifying the effects of short-term rainfall infiltration.

Figure 14e shows the distribution of pore water pressure after 96 h of rainfall, with negative values indicating suction and positive values representing pressure. Rainwater infiltration causes significant changes in pore pressure near the surface, surpassing initial values. A positive pressure zone forms within the saturation zone, while deeper regions exhibit negative pressure fields around 3 × 10⁴ Pa. A high negative pressure zone of 5 × 10⁴ Pa is found between deep and shallow regions. Figure 14f presents the Darcy velocity field, highlighting fluid velocity variations ranging from 2 × 10⁻⁶ m/s to 20 × 10⁻⁶ m/s, influenced by differences in permeability and porosity among overlying strata. The variation in pore water pressure significantly impacts slope stability. As rainfall increases, the pore water pressure within the geomaterial gradually rises, particularly at the top and surface of the slope, where moisture accumulation is more pronounced. This increase in pressure reduces the effective stress within the geomaterial. When the pore water pressure reaches a certain threshold, the effective stress is significantly diminished, thereby weakening the shear strength of the soil and making the slope more susceptible to sliding or failure.

5. Discussion

This case study investigates the deformation of overburden, rainfall runoff, and infiltration due to coal mining activities beneath gully terrain. The findings reveal that mining shallow coal seams in gully regions during flood seasons can trigger a long-term geological disaster chain, driven by multiple factors. The study of gully disaster chains must address both the triggering mechanisms of primary disasters and the cascading mechanisms of secondary disasters. The subsidence of overburden caused by underground coal mining acts as the initiating factor in the gully disaster chain. Based on key stratum theory and mechanical analysis of overlying strata, Sun et al. [53] proposed the Approximate Hyperbolic Subsidence Model to describe the horizontal movement and failure partitioning characteristics of overburden. This paper investigates the horizontal partitioning characteristics of overburden in gully areas, taking into account the actual mining conditions of coal seams. Compared to vertical failure partitioning characteristics, studying horizontal failure characteristics in gully areas is more practically significant. In compression subsidence zones, extensive tensile cracks on slopes intercept rainfall and surface runoff, necessitating timely backfilling by mining technicians. The two-dimensional limit equilibrium method is widely used for slope stability analysis [31,55]. The calculation of slope stability in mining subsidence areas requires the weakening of shear strength parameters based on overburden failure partitioning, and this study provides a basis for such partitioning. Mining depth, mining thickness, and repeated mining activities significantly impact the degree of surface subsidence and are fundamental causes of intense strata movement [48,56,57]. Therefore, sensitivity analysis of these mining parameters and conditions on strata movement characteristics in gully areas is a future research direction. Barrier lakes and runoff infiltration are significant secondary disasters in gully disaster chains. Existing studies have insufficiently explored rainfall-runoff convergence in mining subsidence areas in gully regions. This paper analyzes the rainfall-induced disaster process in the study area through a real rainfall event. The formation of barrier lakes in gully areas is closely linked to landslides; therefore, future mapping of landslide susceptibility in gully areas is an important preliminary task, guiding mine decision-makers in timely mitigation of landslide hazards. The Richards equation is an effective tool for simulating unsaturated flow conditions under specific circumstances [58]. Our research primarily focuses on the initial infiltration process, where unsaturated flow predominates. Although this situation is similar to unsaturated flow conditions in soil, the Richards equation provides a good approximation. However, future work could involve using discrete fracture network (DFN) models or coupled continuum and discrete modeling approaches to refine these simulations further.

While this paper focuses on the quantitative analysis of this disaster chain, it is equally important to develop strategies to break the chain within subsidence areas of gully regions (Figure 15). The disaster chain includes three stages: causation, excitation, and destruction. In the causation stage, mining activity is the primary driver. To minimize overburden subsidence, techniques such as height-restricted mining [59], backfill mining [60], and room-and-pillar mining [61] can be employed. In fully caving operations, goaf grouting may help disrupt the disaster chain [62]. During the excitation stage, unstable slopes and rainfall are the main contributors. Depending on slope conditions, the disaster chain can be interrupted by using geological anchoring technology [63] or ecological slope protection [64]. Traditional rigid concrete ditches may be destabilized by residual ground settlement, so flexible intercepting ditches along the slope crests of the gully are recommended. In the destruction stage, repairing surface cracks in shear subsidence zones and implementing anti-seepage measures in river channels are critical. In assessing geological disaster susceptibility, collapse areas in goafs are rarely considered in landslide susceptibility mapping (LSM) [65,66], and most LSMs focus only on rainfall intensity without accounting for rainfall-runoff inundation. By analyzing rainfall inundation in gully regions, the depth of inundation under varying intensities can be determined. Future research could integrate rainfall-runoff maps with LSM to identify potential areas for barrier lake formation. This could also help mining enterprises set early warning thresholds for rainfall intensity, improving disaster mitigation efforts.

6. Conclusions

The three factors of shallow coal seams, gully terrain, and flood season rainfall present challenges to safe and efficient coal mining. This paper analyzes the overlying rock damage from mining in gully areas, as well as rainfall-induced flooding and infiltration, using on-site observation and numerical simulations. The main conclusions are as follows:

(1)

After shallow coal seam mining, the overlying strata are divided vertically into two zones: caved and fractured zones. The fractured zone is further divided horizontally into compression and shear subsidence zones. These zones are influenced by factors such as the thickness and strength of the key strata (KS) and the gully slope. The shear subsidence zone, which generates both compression and shear deformation, intercepts rainfall runoff and is a critical area for crack landfill treatment.

(2)

The HEC-RAS program efficiently calculates hydrological and hydrodynamic features, such as the water level of the barrier lake (ranging from 4.0 to 8.6 m after rainfall), the maximum flow velocity of the river channel (0.17 m/s), and the inundation range in the study area. By integrating annual maximum daily rainfall data under different frequencies, this program can develop a dynamic risk assessment method for mining-induced geohazards, such as landslides and barrier lakes in gully regions. The combination of LSM and rainfall-runoff simulation shows promise for assessing surface water-related geohazards in mining areas.

(3)

Infiltration and seepage analysis in the gully area shows that under a rainfall intensity of 3.67 mm/h, saturation increases more in low-lying areas, with the infiltration line descending. Seepage prevention measures should be implemented in riverbeds affected by mining-induced subsidence to ensure the sustainability of coal extraction and surface water resources. Matric suction is a critical factor in large-scale numerical modeling. Rainfall infiltration reduces matric suction in unsaturated zones, decreasing the effective stress of rock strata, which in turn reduces their shear strength—a key factor in rainfall-induced landslides.