Analysis of Surface Runoff and Ponding Infiltration Patterns Induced by Underground Block Caving Mining—A Case Study

Abstract

Surface subsidence induced by underground mining in mining areas significantly alters surface topography and hydrogeological conditions, forming depressions and fissures, thereby affecting regional runoff-ponding processes and groundwater infiltration patterns. Accurate assessment of infiltration volumes in subsidence zones under heavy rainfall is crucial for designing underground drainage systems and evaluating water-inrush risks in open-pit to underground transition mines. Taking the surface subsidence area of the Dahongshan Iron Mine as a case study, this paper proposes a rainfall infiltration calculation method based on the precise delineation of surface ponding-infiltration zones. By numerically simulating the subsidence range, the study divides the area into two distinct infiltration characteristic zones under different mining states: the caved zone and the water-conducting fracture zone. The rainfall infiltration volume under storm conditions was calculated separately for each zone. The results indicate that high-intensity mining-induced subsidence leads to a nonlinear surge in stormwater infiltration, primarily due to the significant expansion of the highly permeable caved zone. The core mechanism lies in the area expansion of the caved zone as a rapid infiltration channel, which dominates the overall infiltration capacity multiplication. These findings provide a scientific basis for the design of mine drainage systems and the prevention of water-inrush disasters.

1. Introduction

Mineral resources play a vital role in modern industry and infrastructure development. However, with economic growth and increasing resource demand, the exploitation of shallow deposits has become insufficient, driving mining operations toward deeper strata and more complex geological conditions [1]. This transition presents multifaceted challenges [2]. Mining-induced disturbances cause overlying strata to deform, fracture, and collapse under redistributed stress, leading to surface cracking and subsidence while drastically reducing the integrity of surrounding rock [3]. The resulting hydraulically conductive pathways connect subsurface voids with the surface, creating preferential channels for heavy rainfall, surface runoff, and groundwater infiltration, significantly increasing water-inrush risks [4]. During intense rainy seasons, infiltrated precipitation sharply raises underground water inflows, frequently triggering water inrushes, mine flooding, and even subsurface debris flows [5,6], posing severe threats to personnel safety and infrastructure. Consequently, analyzing rainfall infiltration mechanisms in underground mines and quantitatively assessing infiltration rates are essential for evaluating drainage system capacity and water-inrush hazards. These measures form the scientific foundation for disaster prevention in deep mining operations.

The impacts of rainfall and mining activities are mainly reflected in the weakening of rock mass strength. On one hand, precipitation erosion and pore water pressure reduce the mechanical parameters of structural planes; on the other hand, mining disturbances cause penetrating fractures in surrounding rock. These two factors jointly accelerate the deterioration process of rock mass [7,8,9,10,11], especially in the transition from open-pit to underground mining where multiple stress redistributions occur [12]. This strength attenuation directly changes groundwater migration conditions. As fracture networks expand and connect, the permeability of rock mass shows anisotropic enhancement, transforming rainwater infiltration patterns from homogeneous dispersion flow to fracture-dominated preferential flow [13]. Demand et al. [14] confirmed the complexity of preferential flow through fractures using a large-scale soil moisture sensor network, yet existing models remain insufficient in capturing such dynamic variations. Therefore, revealing the dynamic characteristics of rainfall infiltration under damaged rock mass conditions becomes a key prerequisite for disaster prevention. In terms of theoretical calculations, Zhan et al. [15] derived an analytical solution for unsaturated flow in infinitely long, overburdened layers and studied the influence of rainfall patterns on infiltration processes and deep percolation. Feng et al. [16] proposed multiple exponential functions to describe highly nonlinear soil water retention curves and hydraulic conductivity functions, deriving a semi-analytical solution for pore water pressure in steady-state unsaturated infinite slopes. Wu et al. [17] combined mass conservation, Darcy’s law, and elastic theory to derive an analytical solution for coupled seepage-deformation in unsaturated soil slopes during rainfall infiltration using Fourier integral transform methods. Tracy and Vahedifard [18] proposed an analytical solution for water-mechanics coupling in unsaturated soil infiltration, which accounts for the synergistic effects of pore water pressure and soil deformation. However, this solution is only applicable to one-dimensional homogeneous media and cannot be extended to three-dimensional heterogeneous deformation zones induced by mining. Specifically, it fails to characterize the distinct permeability characteristics between the caved zone and the water-conducting fractured zone, highlighting the limitations of theoretical analytical approaches in complex engineering scenarios. For numerical simulations, Su et al. [19] considered the coupling effects of seepage and stress fields to conduct finite element simulations of seepage and dam slope stability under changing water levels and rainfall intensities based on unsaturated seepage theory. Hou et al. [20] used finite element methods to simulate transient water transport in slopes with different fracture depths. Paronuzzi et al. [21] established numerical infiltration models to study rainfall infiltration processes in homogeneous and stratified colluvial deposits caused by heavy rainfall. Hou et al. [22] used numerical simulation software to study the phenomenon where rainwater initially infiltrates preferentially along fractures in unsaturated fractured soil and then flows away as surface runoff when rainfall exceeds the infiltration capacity. Wu et al. [5] employed COMSOL to solve coupled partial differential equations for soil slopes under rainfall infiltration based on multi-field coupling theory. Zhong et al. [23] proposed a two-step method to accurately simulate coupled surface–subsurface flood processes. Aguilar-López et al. [24] enhanced the dual-permeability model by reconstructing the parameters of the Mualem-van Genuchten soil water retention curve and modifying the hydraulic conductivity exchange function between the matrix and fracture domains. This improvement enabled accurate simulation of preferential flow in fractured clay, demonstrating the model’s applicability in scenarios involving non-capillary flow through wide fractures. Current research mainly focuses on water inflow mechanisms caused by rainfall infiltration, where accurate calculation of infiltration volume is a critical component. For instance, Ma et al. [25] demonstrated in their waste rock pile leaching simulations that dual-permeability models must account for moisture–solute exchange between preferential flow and matrix flow domains. However, their applicability to mining subsidence areas remains limited, as they cannot quantify either the catchment areas induced by surface collapse or the differential permeability characteristics across distinct subsidence zones. Meanwhile, Parsasadr et al. [26] employed a finite element model successfully captured the abrupt permeability enhancement in mining-induced relaxation zones yet failed to establish quantitative relationships between varying degrees of collapse deformation and the spatial heterogeneity of hydraulic conductivity. Further corroborating these findings, Li et al. [13] revealed through coupled rainfall-mining studies that vertically oriented mining-induced fractures synergistically interact with elevated pore water pressure from rainfall infiltration to exacerbate slope instability. Nevertheless, current modeling frameworks still lack precise characterization of the dynamic evolution of permeability properties in deformation zones.

In light of this, as shown in Figure 1, this study proposes a rainfall infiltration calculation method based on the delineation of surface runoff infiltration zones. Taking the Dahongshan Iron Mine as a case study, we established a refined three-dimensional computational model. By integrating field investigations and numerical simulations, we assessed surface deformation characteristics, divided the study area into caved zones and water-conducting fracture zones, and delineated the catchment boundaries surrounding the subsidence area. Different permeability coefficients were assigned to distinct zones, and the open-source hydraulic modeling software Hydrologic Engineering Center’s River Analysis System (HEC-RAS) was employed to calculate rainfall infiltration volumes under storm conditions for each subregion. Through comparative analysis, we examined how mining activities influence rainfall infiltration patterns in subsidence zones at varying mining intensities. The findings provide a scientific basis for designing mine drainage systems and preventing water inrush disasters.

2. Calculation Method of Rainfall Infiltration Based on Infiltration Zone Division

2.1. Fundamental Principles

Currently, multiple methods are available for seepage analysis, each with its own merits and limitations. For instance, MODFLOW demonstrates excellent performance in groundwater flow simulations but requires significant computational resources [27]. The SWAT model is particularly suitable for watershed-scale analyses, though its spatial resolution remains constrained [28]. While MIKE SHE provides comprehensive hydrological modeling capabilities, its implementation demands relatively high-quality input data [29]. HEC-RAS has been demonstrated to exhibit strong relevance in assessing water inrush risks and reliably predicting future flood scenarios within the study area [30,31]. This modeling system maintains balanced performance between computational accuracy and efficiency, characterized by high computational productivity and low parameter sensitivity. The numerical computations of the two-dimensional hydrodynamic model combine the finite volume method and finite difference method, enabling effective calculation of the two-dimensional Saint-Venant equations or diffusive wave equations. The program preprocesses the divided computational grids into cells with a series of hydraulic characteristic curves derived from spatial data, thereby determining the elevation–volume curve for each cell. When the calculated water volume within a cell exceeds the cell’s capacity, overflow occurs. The computational grids are unstructured: grids not adjacent to boundaries are regular quadrilaterals, while those intersecting boundaries adopt irregular polygons (with up to eight sides) to accommodate complex boundaries. The grids support local refinement and alignment. External boundary conditions include water level, flow rate, stage–discharge relationship, and normal depth (channel bed slope). Internal boundaries support flow rate conditions, while initial conditions allow for either a uniform water level or completely dry riverbed. The hydrodynamic analysis mode in the software enables rainfall impact analysis on watersheds within the study area, performing stable and rapid flood inundation simulations. The unsteady flow simulation employs the continuity Equation (1) and momentum Equation (2) as its computational formulas.

$${\displaystyle \frac{\partial h}{\partial t}}+{\displaystyle \frac{\partial \left(hu\right)}{\partial x}}+{\displaystyle \frac{\partial \left(hv\right)}{\partial y}}=0$$

(1)

In the formula, h is the water depth, m; u and v are the flow velocities of the water in the x and y directions, respectively, m/s; t is time, s.

Momentum equation in the x-direction is as follows:

$${\displaystyle \frac{\partial \left(hu\right)}{\partial t}}+{\displaystyle \frac{\partial \left(h{u}^2+0.5g{h}^2\right)}{\partial x}}+{\displaystyle \frac{\partial \left(huv\right)}{\partial y}}={\displaystyle \frac{-gh\partial z}{\partial x}}-{\tau }_x$$

(2)

Momentum equation in the y-direction is as follows:

$${\displaystyle \frac{\partial \left(hv\right)}{\partial t}}+{\displaystyle \frac{\partial \left(h{v}^2+0.5g{h}^2\right)}{\partial y}}+{\displaystyle \frac{\partial \left(huv\right)}{\partial x}}={\displaystyle \frac{-gh\partial z}{\partial y}}-{\tau }_y$$

(3)

In the formula, τ_x, τ_y are the frictional forces at the bottom of the riverbed, N.

Manning’s formula (Equation (4)) runs through the entire hydraulic calculation and is used to calculate water flow velocity, flow rate, frictional slope, etc.

$$V={\displaystyle \frac{1}{n}}{R}^{2/3}{S}^{1/2}$$

(4)

In the formula, V is the average flow velocity, m/s; n is the Manning coefficient, which comprehensively reflects the influence of ground roughness on water flow; R is the hydraulic radius (ratio of wetted perimeter to cross-sectional area); S is the hydraulic slope.

Currently, there is no well-established or mature calculation method for determining rainfall infiltration volume in subsidence areas. Generally, parameter coefficient methods specified in relevant regulations are adopted for estimation purposes. The infiltration volume calculation formula is as follows:

$${\mathrm{Q}}_0{=\mathrm{W}}_0\times \mathsf{\alpha }$$

(5)

$${\mathrm{W}}_0{=\mathrm{H}}_{\mathrm{P}}\times \mathrm{F}/1000$$

(6)

In the formula, Q₀ is the total amount of runoff and seepage in the collapsed area, m³/d; W₀ is the regional surface runoff, m³/d; H_p is the 24 h rainstorm of design frequency, mm; F is the area of surface subsidence area, m²; α is the regional infiltration coefficient.

2.2. Improvement Methods

The rationality of delineating the subsidence area directly affects the accuracy of rainfall infiltration estimation. Conventional methods for determining the subsidence area involve defining the corresponding surface subsidence zone based on the angle of draw and the mining extent of the ore body and then using this area as the basis for estimating rainfall infiltration. However, as mining depths increase, the presence of massive overburden strata leads to distinct characteristics in surface subsidence patterns and ground pressure behavior that differ markedly from shallow underground mining scenarios. Conventional methods employing linear fault angle assumptions—where the angle extends directly from the mining boundary to the surface—fail to accurately capture the nonlinear strata movement mechanisms induced by deep mining. This limitation frequently results in significant deviations in subsidence zone predictions [32]. According to literature [33], surface subsidence caused by underground mining can be divided into three typical zones: (1) caved zone: the rock in this zone completely loses its original structure, texture, and physical-mechanical properties, collapsing into irregular or quasi-stratified rock blocks toward the goaf; (2) water-conducting fracture zone: located between the caved zone and the continuous deformation zone, this area exhibits new fractures and bed separation in the rock layers; and (3) continuous deformation zone: due to reduced support from underlying strata, the rock layers undergo plastic bending or slow, uniform subsidence under their own weight. The caved zone contains numerous interconnected voids between rock blocks, significantly enhancing permeability and serving as an excellent channel for rainfall infiltration into the mine workings. Thus, the height and extent of the caved zone directly influence atmospheric precipitation infiltration into the mine. The water-conducting fracture zone also exhibits increased permeability due to rock deformation and enhanced fracture development. If these fractures extend to the surface, they can become effective pathways for precipitation infiltration. In contrast, the continuous deformation zone experiences minimal changes in rock structure and seepage capacity and is generally not considered in infiltration assessments.

This study assumes that rainfall infiltration in the subsidence area comes from two sources: (1) infiltration directly through the caved zone and (2) infiltration directly through the water-conducting fracture zone. The permeability characteristics of rock masses undergo varying degrees of alteration depending on the magnitude of subsidence deformation. It is therefore inappropriate to apply a uniform rainfall infiltration coefficient across the entire subsidence area. The methodology for calculating rainfall infiltration presented in Reference [34] addresses this issue by the following:

$${\mathrm{Q}=\mathrm{W}}_1\times {\mathsf{\alpha }}_1{+\mathrm{W}}_2\times {\mathsf{\alpha }}_2\times \mathsf{\Phi }$$

(7)

In the formula, Q is the total amount of runoff and infiltration in the collapse area during rainstorm at the design frequency, m³/d; W₁ is the area of the falling zone, m², defined based on numerical simulation results; α₁ is the infiltration coefficient of the falling zone area when the design frequency is rainstorm; W₂ is the area of the infiltration directly through the water-conducting fracture zone, m², defined based on numerical simulation results; α₂ is the infiltration coefficient of the water-conducting fissure zone area when the design frequency is rainstorm; Φ is the surface runoff coefficient, a dimensionless parameter describing the efficiency of converting rainfall into surface runoff, which is the proportion of surface runoff in a certain period of time.

3. Surface Deformation Prediction and Infiltration Zone Division Based on Numerical Simulation

3.1. Engineering Background

The Dahongshan Iron Mine is located in Gasa Town, Xinping Yi and Dai Autonomous County, Yuxi City, Yunnan Province, with geographical coordinates of 101°39′ east longitude and 24°06′ north latitude. It is situated on the eastern bank of the Gasa River (the upper reaches of the Red River and Yuan River), adjacent to the eastern side of the Ailao Mountain Range. The surface elevation of the mining area ranges from 600 m to 1850 m, characterized by an eroded and denuded mountainous terrain with deep incisions, significant undulations, and well-developed reticulated valleys. The Manggang River, Feiwei River, and Laochang River flow through the mining area, converging into the Hunlong River, which joins the Gasa River approximately 9 km southwest of the mining area. The Dahongshan Iron Mine is located in a subtropical climate zone with distinct dry and wet seasons. The region receives abundant rainfall, with an annual precipitation of 700–1200 mm (average 930.8 mm). Over 60% of the rainfall occurs between June and September, primarily in the form of showers and heavy rainstorms.

At present, the Dahongshan Iron Mine is simultaneously conducting mining operations across multiple ore bodies within an area approximately 2000 m long east–west and 1000 m wide north–south, at elevations ranging from +360 m to +1300 m. The location of the mining area is shown in Figure 2. These operations include open-pit mining of shallow lava iron ore, deep mining of the II1 iron ore group, underground mining of the I copper ore belt, and the III–IV ore groups. This has resulted in a globally unique mining scenario featuring combined surface-underground extraction, concurrent shallow and deep mining, and the simultaneous application of three mining methods: caving, open stope, and backfill. The operation has established a large-scale, multi-section, multi-district concurrent mining pattern.

3.2. Establishment of Three-Dimensional Refined Model

Precision modeling can accurately reconstruct orebody morphology and geological structures (such as faults, fractures, and weak interlayers), providing a computational domain foundation for numerical simulations that more closely approximates real-world conditions. The Dahongshan Iron Mine features complex lithological distribution, with main rock types including gabbro-diabase, metasomatic sodium lava, dolomitic marble, copper ore belts, and iron ore belts. The area contains F1 and F2 fault structures and fold structures, as well as shallow and deep ore bodies. The F1 fault strikes northwest–west or nearly east–west, dips south or southwest with an inclination angle of 60–85°, and has a displacement exceeding 500 m. The F2 fault is inferred to have initially been a reverse fault and later transformed into a normal fault. It strikes nearly east–west, extends over 1.1 km, dips south with an inclination angle of approximately 80°. Both fault zones are filled with intruded gabbro-diabase. The F2 fault serves as the natural boundary between shallow and deep ore bodies, while the F1 fault marks the southern boundary of the deep ore bodies.

To mitigate and eliminate boundary effects in large-scale simulation calculations [25], the geometric model boundaries were extended by 100 m beyond the projected areas of both open-pit and underground mining zones. The model was constructed at a 1:1 scale, with initial dimensions of 3000 m × 2500 m × 1300 m (length × width × height, where the Z-axis elevation varies according to topographic contours). To ensure modeling precision, the absolute tolerance was set at 0.01 and the angular tolerance at 1°.

Research indicates that topography significantly influences surface subsidence, necessitating accurate representation of surface relief in computational models [35,36,37]. The mine employed unmanned aerial-vehicle oblique photogrammetry and laser scanning technology to acquire high-resolution surface point cloud data, which was processed to generate a precise 3D surface entity model as shown in Figure 3.

Based on the mining schedule, yearly excavation stope models were established to serve as the foundation for subsequent analysis and calculations, as illustrated in Figure 4.

The geometric modeling of different lithologies was achieved through cross-section profiling and block cutting methods. A specific elevation plane containing multiple section lines was selected as the reference surface for placing various cross-sections. Each section bearing lithological boundaries was sequentially positioned on this reference plane. The modeling accuracy improves with increasing number of incorporated cross-sections. For this study, 18 exploration profiles (A28–A45) with 50 m horizontal spacing between adjacent sections were utilized. Contour lines of identical lithologies at corresponding positions across all sections were identified to determine the spatial configuration and positioning of each rock formation. These contours were then integrated to form three-dimensional lithological interfaces. The base geometric model was subsequently partitioned using these established rock mass interfaces to obtain separate geometric models for each lithology. Based on this framework, the caved zone body was constructed according to field-measured vertical collapse extents. The entire modeling process strictly adhered to geological profiles obtained through precise mine exploration, achieving accurate representation of lithological distributions in the Dahongshan Iron Mine, as demonstrated in Figure 5.

3.3. Range of Surface Movement Induced by Mining and Division of Infiltration Zones

The rock mechanical parameters for various lithologies in the Dahongshan Iron Mine were obtained through laboratory tests. Combined with structural plane joint identification for rock quality reduction [38], the Hoek–Brown empirical estimation method was employed to calculate the mechanical strength of rock masses. The final parameters used for numerical simulation are shown in

Table 1. Numerical simulation of rock mechanics parameters of Dahongshan Iron Mine.

Iithology	Natural Unit Weight (g/cm3)	Elastic Modulus (GPa)	Poisson’s Ratio	Tensile Strength (MPa)	Cohesion (MPa)	Friction Angle (°)
Falling objects	2.10	12.30	0.17	0.50	0.50	38.0
Material	1.40	12.30	0.45	0.01	0.01	40.0
Foundation	1.81	0.60	0.33	0.10	0.20	43.2
Dolomite albitite	3.12	17.94	0.28	2.96	2.50	45.4
Gabbro diabase	2.80	16.42	0.31	2.75	2.75	46.8
Sodium lava	2.81	19.51	0.30	4.03	2.67	45.8
Dolomite marble	2.86	18.82	0.31	3.84	2.59	46.6
Iron ore	3.60	21.33	0.28	3.54	2.85	48.4
Copper mine	3.50	16.42	0.27	2.76	2.75	46.8

.

The initial stress field adopted for this numerical calculation is characterized as follows: The maximum principal stress in the mining area is oriented along the NNE-SSW direction, ranging between N16°E and N65.68°E, with an average orientation of N40.84°E. All principal stresses generally increase with depth, exhibiting the following variation patterns:

$$\begin{array}{l}{\sigma }_1=1.337+0.047H\\ {\sigma }_2=\gamma H\\ {\sigma }_3=0.59+0.22H\end{array}$$

(8)

In the formula, ${\sigma }_1$ is the maximum principal stress, MPa; ${\sigma }_2$ is the intermediate principal stress, MPa; ${\sigma }_3$ is the minimum principal stress, MPa; H is the depth of the measuring point, m.

The open-pit mining is expected to be completed by 2026 and will transition entirely to underground mining. Accordingly, the numerical simulation analysis considers three working conditions based on different underground mining extents: (1) current status; (2) mining until 2026; and (3) full extraction.

For the computational domain boundaries, the (1) initial geostress is first applied. (2) Displacement constraints are then implemented: All nodes at the model bottom are constrained in the x, y, and z directions. Both ends in the x-direction are constrained along x; both ends in the y-direction are constrained along y; the model top remains a free boundary [39]. The Mohr–Coulomb elastoplastic constitutive model is adopted for the simulation. The numerical computation process follows these steps: Initial geostress equilibrium under unmined conditions, with maximum unbalanced force set to 1.0 × 10⁻⁵ and planned stope-by-stope excavation calculations until final equilibrium is achieved.

A comparative analysis was conducted between the cumulative deformation data obtained from eight total stations installed around the subsidence crater and the numerical simulation results, as presented in

Table 2. Comparison between numerical simulation results and monitoring data.

Point	Actual Displacement Changes	Numerical Simulation Results	Error
G1	−28.3 mm	−25.58 mm	−9.6%
G5	−21.3 mm	−21.77 mm	2.2%
G6	−22.4 mm	−24.19 mm	7.9%
G7s3	−22.6 mm	−24.72 mm	9.4%
G8s2	−24.6 mm	−26.33 mm	7.0%
G9	−24.1 mm	−25.92 mm	7.5%
G15s2	−28.5 mm	−26.51 mm	−6.9%
G22	−29.0 mm	−27.98 mm	3.5%

. The discrepancy between field monitoring data and numerical calculations was found to be within 10%, demonstrating the high reliability of the selected mechanical parameters and refined modeling approach employed in the numerical simulations.

Based on numerical simulation results, surface deformation data under different working conditions were obtained. The caved zone and water-conducting fracture zone were delineated according to critical deformation values specified in technical standards (incline i in mm/m, curvature k in 10⁻³/m, and horizontal deformation ε in mm/m). The critical deformation thresholds are defined as follows:

Caved zone: incline (i): ±10 mm/m; curvature (k): ±0.6 × 10⁻³/m; and horizontal deformation (ε): ±6 mm/m;

Water-conducting fracture zone: incline (i): ±3 mm/m; curvature (k): ±0.2 × 10⁻³/m; and horizontal deformation (ε): ±2 mm/m.

The resulting surface contour maps for all three working conditions are presented in Figure 6, Figure 7 and Figure 8.

4. Analysis of Surface Runoff-Ponding Infiltration Patterns Induced by Mining Activities

4.1. Simulation of Surface Runoff in Mining Areas Based on HEC-RAS

The high-precision surface DEM model for Working Condition 1 was generated by converting high-accuracy point cloud data of the study area obtained through unmanned aerial-vehicle oblique photogrammetry, with the imported topographic map shown in Figure 9a. The final surface deformation results from numerical simulations were exported to create surface DEM models for Working Conditions 2 and 3, as presented in Figure 10a and Figure 11a. The computational domain was delineated along the river channels, with a 2D grid spatial resolution of 10 m, resulting in a total of 88,358 grid elements. The generated meshes are displayed in Figure 9b, Figure 10b and Figure 11b.

After mesh generation, appropriate boundary conditions must be configured according to actual conditions. Boundary conditions primarily define the inflow and outflow conditions of the computational domain, while initial conditions specify the starting state of the entire calculation area. The model’s external boundaries mainly include settings along the outer perimeter of the 2D model (water level, discharge, stage–discharge relationship, channel bed slope, and precipitation conditions and water levels distributed across the entire 2D grid). The numerical model was configured with all external boundaries set as flow outlets under a hydraulic gradient of 0.03. The global rainfall condition was applied as the inflow boundary, incorporating both actual rainfall intensities over a two-month period and a 5-day continuous precipitation event equivalent to a 50-year storm recurrence interval (the rainfall input curve is shown in Figure 12). The simulation adopted a 15 min computational time step with synchronized output intervals for hydraulic parameters. A uniform Manning’s roughness coefficient of 0.05 was assigned throughout the study area to represent the hydraulically smooth characteristics of the fractured bedrock surface. The simulation results demonstrating the system’s hydrological response are presented in Figure 13.

4.2. Estimation of Rainfall Infiltration in Mining Area Under Rainstorm Condition

Based on the results, the catchment area surrounding the subsidence zone was delineated, and the final computational domain is shown in Figure 14. In the figure, the blue line represents the boundary of the caved zone, while the yellow line indicates the extent of the water-conducting fracture zone.

The infiltration coefficient and runoff coefficient are critical parameters affecting the accuracy of rainfall infiltration estimation, and different infiltration coefficients should be applied to different zoned areas. In accordance with the relevant provisions of the “Code for Design of Metal Mine” (GB 50830-2013) [40] (

Table 3. Reference values for stormwater infiltration coefficients for avalanche areas.

Degree and Characteristics of Rock (Soil) Layer Damage on Collapsed Surfaces and Ore Body Roof	Characteristics of Overlying Soil on the Ore Body		Design-Frequency Rainstorm Infiltration Coefficient
The collapse zone has not extended to the surface; only the water-conducting fracture zone has extended to the surface.	Nonplastic waterproof soil layer	Brittle rock	0.20~0.15
		Plastic rock	0.15~0.10
	Plastic waterproof soil layer Thickness (m)	5~10	0.10~0.05
		11~20	≤0.05
The overlying rock at the top of the ore body does not collapse repeatedly.	Nonplastic waterproof soil layer	Brittle rock	0.35~0.30
		Plastic rock	0.30~0.20
	Plastic waterproof soil layer Thickness (m)	5~10	0.20~0.15
		11~20	0.15~0.10
		21~30	0.10~0.05
		31~50	≤0.05
Repeated collapse of overlying rock at the top of the ore body	Nonplastic waterproof soil layer	Brittle rock	0.40~0.30
		Plastic rock	0.30~0.25
	Plastic waterproof soil layer Thickness (m)	5~10	0.25~0.20
		11~20	0.20~0.15
		21~30	0.15~0.10
		31~50	0.10~0.05

(1) Plastic rocks in the table generally refer to shale, mudstone, argillaceous sandstone, tuff, phyllite, etc; brittle rocks generally refer to limestone, dolomite, marble, granite, gneiss, diorite, etc; Plastic waterproof soil layer refers to the bedrock composed of Quaternary clay, sub clay, and severely weathered soil-like substances. (2) For the fluctuation value of the rainstorm infiltration coefficient of design frequency in the table, the smaller value is taken when the depth ratio is large; when the depth ratio is small, the maximum value should be taken when the water-conducting cracks or collapse zones first reach the surface.

), the study area features overburden rock above the ore body characterized by repetitive collapse and brittle rock strata lacking a plastic impermeable layer. The recommended infiltration coefficient for design-frequency heavy rainfall is suggested to be 0.4–0.3. For the caved zone, this region is significantly affected by repeated mining disturbances, exhibiting a high degree of rock fragmentation and lacking cohesive backfill to impede infiltration. Consequently, it demonstrates a greater infiltration capacity under heavy rainfall conditions, justifying the adoption of the upper limit value of 0.4 as recommended by the standards. In contrast, the water-conducting fractured zone, while containing a certain degree of fracture networks, is located in a transitional area between the caved zone and intact rock strata, resulting in reduced effective infiltration pathways. Hence, a more conservative value of 0.3 is selected. This differential treatment better aligns with the field-observed seepage mechanisms—where pipe flow dominates in the caved zone, while fracture flow prevails in the water-conducting fractured zone—compared to traditional, uniform coefficient methods. It more accurately reflects the actual infiltration characteristics of each partition under heavy rainfall conditions. As for the runoff coefficient, in accordance with the provisions of the “Code for Design of Outdoor Wastewater Engineering” (GB 50014-2006) [41] (

Table 4. Table of runoff coefficient values.

Ground Type	Runoff Coefficient Φ
Various roofs or concrete or asphalt pavements	0.85~0.95
Gravel pavement with large stone paving or asphalt surface treatment	0.55~0.65
Graded crushed stone pavement	0.40~0.50
Dry masonry or gravel pavement	0.35~0.40
Unpaved soil pavement	0.25~0.35
Park or green space	0.10~0.20

), a value of 0.3 is adopted for unpaved soil surfaces.

The calculation results were imported into ArcGIS 9.3 to calculate the water collection in each region, and Equation (6) was used to calculate the rainfall infiltration in each working condition of the mining area. The calculation results are shown in

Table 5. Estimated results of total storm water infiltration (actual rainfall).

Region	Area (m3)	Water-Collecting Amount (m3/d)	Infiltration Coefficient	Runoff Coefficient	Infiltration Capacity (m3/d)	Water Discharge
Caved Zone	946,181.043	9871.772	0.4	-	4820.5816	4590.806
Water-Conducting Fracture Zone	1,404,063.217	3020.809	0.3	0.3

and

Table 6. Estimated results of total storm water infiltration (50-year return period storm intensity).

	Region	Area (m3)	Water-Collecting Amount (m3/d)	Infiltration Coefficient	Runoff Coefficient	Infiltration Capacity (m3/d)
Condition 1	Caved Zone	946,181.043	78,145.303	0.4	-	32,704.6594
	Water-Conducting Fracture Zone	1,404,063.217	16,072.647	0.3	0.3
Condition 2	Caved Zone	1,115,764.036	184,942.049	0.4	-	74,915.6012
	Water-Conducting Fracture Zone	1,475,216.672	10,430.907	0.3	0.3
Condition 3	Caved Zone	1,725,046.769	210,728.485	0.4	-	85,401.1802
	Water-Conducting Fracture Zone	2,152,679.054	12,330.958	0.3	0.3

.

The computational results demonstrate that the discrepancy between the actual monitored drainage volume and the infiltration quantity calculated using the proposed method is 8.06%, indicating a high level of accuracy for this methodology.

4.3. Result Analysis

Due to intensive mining operations at the Dahongshan Iron Mine, numerous collapse pits of varying sizes have formed on the surface. Under rainfall conditions, these pits transform into subsidence lakes, significantly altering both the area and spatial distribution patterns of rainfall infiltration zones. As mining progresses, the subsidence area expands, with particularly notable increases in the caved zone area. As shown in Figure 15, numerical simulations reveal that under Working Condition 2 compared to Condition 1, the infiltration area of the subsidence zone expands by approximately 10%, with newly formed caved zones accounting for over 70% of this increase. The peripheral caved zones and water-conducting fracture zones further extend the influence range of infiltration. Water accumulation areas leak into the goaf through hydrostatic pressure, resulting in a doubling of the infiltration volume. Under Working Condition 3, the subsidence area can reach 3.4 times the projected area of the goaf. Compared to Condition 1, the infiltration area of the subsidence zone expands by approximately 65%, with newly formed caved zones making up over 50% of this expansion. These caved zones become preferential pathways for rapid rainfall infiltration. Water accumulation areas directly leak into the goaf under hydrostatic pressure, increasing the stormwater infiltration volume in the subsidence zone by 2.4 times.

Since caved zones typically exhibit significantly higher permeability than the original ground surface and water-conducting fracture zones, their expansion directly enhances the overall rainfall infiltration capacity. Consequently, under identical storm conditions, the total volume of rainfall infiltrating underground increases substantially. This growth in infiltration volume is not linear but closely correlated with the expansion of highly permeable water-conducting fracture zones. During extreme storm events, the rapid water-conducting capability of caved zones further amplifies their contribution to total infiltration. Therefore, changes in the infiltration zone area due to mining disturbances—particularly the expansion of high-permeability zones—constitute the core mechanism governing increased rainfall infiltration in mining areas. Precisely delineating these zones and quantifying their area changes are critical for accurately assessing the impact of mining on rainfall infiltration.

4.4. Prevention and Control Measures

The surface subsidence area is classified into three hierarchical anti-seepage zones based on deformation characteristics, inundation extent, and temporal factors, with detailed classification criteria and corresponding mitigation measures provided in

Table 7. Table of seepage control classifications and seepage control measures.

Classification of Anti-Seepage Areas	Anti-Seepage Grade	Anti-Seepage Measure
Flooded zones encompassing both the caved zone and water-conducting fracture zone	Level 1	Stratified fracture grouting Geomembrane installation Pumping drainage system
Non-flooded sections of the caved zone	Level 2	Stratified fracture grouting
Non-inundated portions of the water-conducting fracture zone	Level 3	Direct fracture grouting
Supplementary anti-seepage measures	Circular cutoff drainage system installed at the collapse area periphery

. The primary containment zone (Grade I) comprises inundated sections of both the caved zone and water-conducting fracture network, while the secondary zone (Grade II) includes non-flooded portions of the caved zone, and the tertiary zone (Grade III) covers non-submerged areas of the fractured zone. Differential anti-seepage measures are implemented accordingly: For Grades I and II, stratified fracture grouting is employed through sequential borehole injection of cementitious and chemical grouts into bedrock discontinuities under controlled pressure, facilitating permeation and filling of rock/soil fractures to form continuous impermeable barriers upon curing. The primary zone receives supplementary protection via geomembrane installation coupled with active pumping systems for seepage reduction. Grade III areas undergo direct fracture infilling treatments. Complementing these zonal measures, an annular water interception system is constructed along the subsidence perimeter through composite engineering of earthen cofferdams and open drainage channels, achieving controlled diversion of surface runoff and effective hydraulic isolation of the collapse area from external water sources.

5. Conclusions and Discussion

5.1. Conclusions

Our conclusions include the following:

(1)

A rainfall infiltration calculation method based on precise delineation of surface catchment infiltration zones was proposed. By integrating field investigations, numerical simulations, and HEC-RAS catchment modeling, the caved zones and water-conducting fracture zones under different mining stages were identified. Different rainfall infiltration coefficients were applied to various infiltration zones, leading to the derivation of a calculation formula for rainfall infiltration volume in surface subsidence areas.

(2)

A refined 3D mine model was established using 18 exploration profiles to accurately reconstruct the complex geological environment of Dahongshan Iron Mine. Detailed stope models were created according to mining schedules, providing a foundation for simulating realistic mining-induced surface deformation. A high-precision DEM model of Shilu Iron Mine’s surface was developed using UAV oblique photography data, enabling surface catchment simulation under storm conditions and obtaining water accumulation patterns in mining-induced subsidence zones.

(3)

Underground mining reshapes the spatial distribution and infiltration mechanisms through surface deformation. With expanding the mining scope, in Working Condition 2, the infiltration area increased by 10% compared to Condition 1, with over 70% being newly formed caved zones; in Working Condition 3, the infiltration area expanded by 65% versus Condition 1, with newly formed caved zones accounting for over 50%. The peripheral expansion of caved zones and water-conducting fractures doubled the stormwater infiltration volume in Condition 2 and increased it by 2.4 times in Condition 3 compared to Condition 1. This nonlinear growth stems from significantly enhanced overall infiltration capacity due to expanded high-permeability caved zones, which serve as preferential pathways for rapid rainfall percolation and constitute the core mechanism controlling infiltration volume increases.

5.2. Discussion

The discussion is as follows:

(1)

The applicability of the model is constrained by specific geological conditions. When applied to other mining sites, parameter adjustments and validation must be conducted based on their unique geological characteristics. Rock mechanical parameters were determined through a combination of field tests and numerical inversion. However, due to limitations in exploration data density, spatial variability in localized regions may not be fully characterized. While the Mohr–Coulomb criterion effectively captures shear failure in rock masses, it does not account for strength degradation under cyclic loading. Future improvements could incorporate damage mechanics models to address this limitation.

(2)

Given that the waste backfill material in this study primarily consists of coarse-grained particles with large pore sizes, capillary effects exhibit a limited influence, and the difference between saturated and unsaturated permeability coefficients is negligible. Consequently, capillary effects were disregarded. However, for materials containing clay layers or fine-grained tailings, capillary action significantly impacts seepage behavior and must be accounted for using unsaturated flow models such as the Van Genuchten model. Future research should conduct comparative analyses of seepage characteristics across different backfill materials to refine mining seepage theory and develop more universal modeling approaches for varying geological conditions.

(3)

Poroelastic theory and fluid–solid coupling analysis hold substantial value for mechanistic studies. However, for large-scale models, fully coupled poroelastic analysis faces practical challenges, including high computational resource demands and insufficient permeability field data. Future work could focus on small-scale poroelastic simulations in critical zones, combined with parameter inversion using surface monitoring data, to enhance understanding of localized deformation mechanisms. While the current simplified approach provides effective support for engineering decision-making, refined poroelastic modeling represents a promising direction for subsequent research to improve prediction accuracy and model applicability.