3D geological fine modeling and dynamic updating method of fault slope in open-pit coal mine Authors: Jie Yuan, Guangwei Liu, Senlin Chai Published: 2024 ABSTRACT AbstractCombining the requirements for refined modeling and dynamic update of fault slope geological models in open-pit coal mines, we systematically elaborated on the elements and rules of slope 3D geological modeling and proposed a fine modeling and dynamic updating method based on digital elevation model (DEM) and half-edge boundary representation (B-Rep) data structures. Initially, the stratigraphic division of the study area is conducted based on borehole and section data, combined with the geological evolution history and the level set theory. Applying inverse distance weighting (IDW) for estimation, the geological interface triangular irregular network (TIN) is constructed using the TIN algorithm. Secondly, fine reconstruction of fault surfaces through Morphing transformation improves the accuracy and smoothness of fault surface shape reconstruction under sparse data conditions. Then, the intersection-cutting processing method of fault-fault and fault-stratum for slope geological ground state model modeling under complex fault segmentation is investigated. Finally, constructing 3D geological models of fault slopes and the integrated correction of surface and subsurface are realized through dynamic updating methods for 3D surfaces and strata. Taking the 3D geological modeling of a fault slope of an open-pit coal mine in China as an example, the feasibility of the technical method is verified.Keywords: Open-pit coal mine, Fault slope, 3D geological model, Fine modeling, Dynamic updatingSubject terms: Civil engineering, Tectonics, Structural geology, Stratigraphy, Natural hazards Introduction Open-pit mine slopes serve as the safety and economic boundaries of open-pit mining operations and are also the object of production activities, thereby playing a crucial role in ensuring mine safety and production efficiency. Landslides represent the most significant safety hazard in open-pit mining and are among the crucial factors influencing the green and efficient extraction of open-pit mines. In recent years, as open-pit mining continues to expand towards deeper, higher intensity, and more significant scale operations, the number of high and steep slopes has gradually increased. Consequently, the issue of landslide disasters in mines has become increasingly prominent, posing a severe threat to the normal production of mines and the safety of lives and property within mining areas1. Experimental and numerical simulation methods are crucial tools for solving practical engineering problems2–5. They enhance the precision of engineering design while ensuring the safety and efficiency of project implementation. In the process of slope stability analysis and calculation, fine exploration of geological conditions is a fundamental link. Ascertaining and reconstructing the geological conditions of coal mining transparency are essential guarantees for conducting slope engineering research and serve as the prerequisite and basis for assessing potential landslide modes in engineering slopes. Numerous scholars researched 3D geological modeling and visualization in conjunction with their respective fields, proposing a series of modeling theories and techniques. Lemon and Jones6 proposed the HORIZONS modeling method, which integrates borehole and section data to construct 3D solid models. Zhu et al.7 considered complex geological structures such as faults and folds and combined borehole and fault data to construct 3D models containing complex geological phenomena. Zhang et al.8 introduced a 3D stratigraphic modeling method based on massive borehole data. They established a stratigraphic model with strict geometric topological relationships through dynamic search area and regional blocking. Jiao et al.9 extracted stratigraphic boundary information and horizon information based on borehole data and performed spatial interpolation and fitting on the stratigraphic data to achieve the construction of geological models based on multi-layer DEM. Lei et al.10 proposed a method for segmenting and modeling 3D complex geological bodies based on multiple data sets and geological prior knowledge constraints, achieving seamless integration between models and enhancing the modeling efficiency and reliability of geological models. However, current research on 3D geological modeling methods still lacks sufficient description of complex geological structures such as faults, making it difficult to achieve fine 3D modeling of geological models containing complex fault structures. At the same time, slope geological models mainly rely on static models, making it challenging to achieve dynamic updates as mining and excavation progress. This ignores the dynamic characteristics of slopes and reduces the accuracy and reliability of slope engineering stability analysis, directly impacting mining operations’ economic efficiency and safety. Therefore, researching dynamic and fined 3D geological modeling methods for open-pit mine slopes is of significant theoretical and practical importance for ensuring mines’ safe and efficient production. In summary, a comprehensive study was conducted on the modeling elements, modeling rules, data structures, topological relationships, modeling processes, and dynamic updating methods of 3D geological models for slopes, so as to meet the requirements of geological models on dynamic and fined modeling for the slopes containing faults. The feasibility of this method was verified by constructing a geological model of a fault-containing slope of an open-pit coal mine. The modeling results showed that this method can be used to construct a fined 3D geological model suitable for slope stability analysis applications and can complete the model’s quick update and reconstruction to provide a model basis for slope stability analysis and evaluation. Essential elements and rules of slope geological modeling Essential elements 3D geological modeling interprets 3D geological objects by quantifying characteristic information such as internal topological relationships, geometric shapes, and physical properties of 3D geological bodies. Topology, geometry, and attributes are the three essential elements of 3D geological modeling11, and they are also indispensable parts of slope geological modeling. Topology elements are divided into micro-topology and macro-topology12,14, akin to the “skeleton” supporting the structure of the human body, ensuring the stable shape of geometric elements. Micro-topology focuses on the adjacency and correlation relationships among elements such as points, lines, and surfaces within the geometric shapes, reflecting the connection and arrangement of geometric elements in space. This aspect is crucial for accurately expressing geological models and quality control and is often directly stored in corresponding data structures. Macro-topology expresses the macroscopic contact relationships between geological structural elements, providing a detailed description of the spatial relationships between geological surfaces and enabling the model to reflect underground geological structures accurately. The geometric elements describe the spatial position and geometric shape of geological elements themselves, akin to the “muscles” attached to the "skeleton," recording information such as the position, size, dimensions, and shape of the morphological object. The attribute elements record the physical and mechanical properties of the slope geological body itself, including lithology, capacity, cohesion, friction angle, etc. Modeling rules Slope geological models serve the purpose of finite element stability analysis solutions. The established models are usually non-manifold structural models, but the modeling algorithms are often based on manifold entities. Additionally, a qualified and effective geological model involves precise fitting of geological exploration data on the surface and encompasses the correct relationships between geological interfaces. Therefore, it is necessary to establish certain rules for geological model construction to handle object boundaries and the connectivity of object meshes, ensuring the correctness of topological constraints and meeting real geological conditions11,13,14. (1) Effectiveness of geological constraints: 3D geological modeling aims to reflect rock strata’ real stratigraphic positions and geometric shapes. During the modeling process, it is necessary to convert the explored geological data into hard constraints for geological modeling, ensuring that the geological interfaces strictly pass through the corresponding stratigraphic sampling points to guarantee the accuracy and authenticity of the modeling. (1) Effectiveness of geological constraints: 3D geological modeling aims to reflect rock strata’ real stratigraphic positions and geometric shapes. During the modeling process, it is necessary to convert the explored geological data into hard constraints for geological modeling, ensuring that the geological interfaces strictly pass through the corresponding stratigraphic sampling points to guarantee the accuracy and authenticity of the modeling. (2) Effectiveness of geological interfaces: The effectiveness of geological interfaces mainly pertains to micro-topology, ensuring the correct topological relationships among grid points, lines, and surfaces, which form the foundation of 3D geological modeling and applications. Each independent stratigraphic interface can be represented by an oriented manifold grid, where each edge, except for the open edges at the boundary occupied by only one triangular patch, is shared by two triangular patches (Fig. 1e). Moreover, the neighborhood triangular patches of grid vertices constitute a closed or open fan, and the triangular patches should maintain consistent normal directions. Therefore, situations depicted in Fig. 1a–d should not occur during the geological interface modeling process. (2) Effectiveness of geological interfaces: The effectiveness of geological interfaces mainly pertains to micro-topology, ensuring the correct topological relationships among grid points, lines, and surfaces, which form the foundation of 3D geological modeling and applications. Each independent stratigraphic interface can be represented by an oriented manifold grid, where each edge, except for the open edges at the boundary occupied by only one triangular patch, is shared by two triangular patches (Fig. 1e). Moreover, the neighborhood triangular patches of grid vertices constitute a closed or open fan, and the triangular patches should maintain consistent normal directions. Therefore, situations depicted in Fig. 1a–d should not occur during the geological interface modeling process. (3) Effectiveness of macroscopic contact of geological interfaces: The macroscopic contact of geological interfaces describes the interaction relationships between strata and strata, strata and faults, faults and faults, and other geological interfaces. According to the principle of topology, geological interfaces constructed using boundary representations can only intersect at common boundaries, adhering to the fundamental laws of geological succession. For any two strata interfaces, Hi and Hj, Hi should only be located on one side of Hj (Fig. 2b, c), and vice versa14. Overlapping stratigraphic interfaces should be avoided (Fig. 2a). Furthermore, geological interfaces should accurately divide the 3D geological body into mutually independent spaces (Fig. 2e, f), and conditions where different strata are interconnected should not exist (Fig. 2d). (3) Effectiveness of macroscopic contact of geological interfaces: The macroscopic contact of geological interfaces describes the interaction relationships between strata and strata, strata and faults, faults and faults, and other geological interfaces. According to the principle of topology, geological interfaces constructed using boundary representations can only intersect at common boundaries, adhering to the fundamental laws of geological succession. For any two strata interfaces, Hi and Hj, Hi should only be located on one side of Hj (Fig. 2b, c), and vice versa14. Overlapping stratigraphic interfaces should be avoided (Fig. 2a). Furthermore, geological interfaces should accurately divide the 3D geological body into mutually independent spaces (Fig. 2e, f), and conditions where different strata are interconnected should not exist (Fig. 2d). (4) Topological consistency of geological models: The topological consistency of geological models is crucial to ensure the effectiveness of geological interface combinations. It is necessary to ensure geometric topological consistency at the intersections of geological interfaces, where points and edges are shared by geological interfaces (Fig. 3). Correct topological relationships are the basis for Boolean operations and volume element meshing of geological models15. (4) Topological consistency of geological models: The topological consistency of geological models is crucial to ensure the effectiveness of geological interface combinations. It is necessary to ensure geometric topological consistency at the intersections of geological interfaces, where points and edges are shared by geological interfaces (Fig. 3). Correct topological relationships are the basis for Boolean operations and volume element meshing of geological models15. (5) Airtight of geological models: In order to facilitate later editing and modification of models, geological interfaces are often established slightly larger than the modeling scope. Based on the rules above, ensuring the airtightness of models by imposing external boundary constraints on the geological model guarantees its integrity. That is, there are no open edges in the final geological model. (5) Airtight of geological models: In order to facilitate later editing and modification of models, geological interfaces are often established slightly larger than the modeling scope. Based on the rules above, ensuring the airtightness of models by imposing external boundary constraints on the geological model guarantees its integrity. That is, there are no open edges in the final geological model. Effectiveness of geological interface grid construction. Effectiveness of macroscopic contact of geological interfaces (Modified from reference [14]). Topological consistency of geological interfaces. 3D data structure The 3D data structure organizes spatial data according to specific rules to accurately reflect spatial entities’ geometric shapes and topological relationships, which is the basis for 3D slope geological modeling. The open-pit mine slope geological model integrates the current topographic structure of the mine stope, the coal and rock geological body structure, and the geological structural structure. The model has high complexity, a large amount of modeling data, and high requirements for modeling accuracy and visual analysis. Therefore, selecting suitable 3D data structures and establishing correct topological relationships are essential to efficiently modeling and dynamically updating complex slope geological models. By analyzing commonly used 3D data structures and combining the advantages of DEM and B-Rep half-edge data structures, the 3D data structure for open-pit coal mine slopes containing faults and other complex geological structures can be established. In this structure, the DEM is used as the boundary curves of the slope geological model and as surfaces for clipping geological strata. The half-edge structure of B-Rep solid data is applied to organize the spatial topological relationships of DEM, thereby achieving an informative description of complex open-pit coal mine slope geological models. Overview of DEM The stratum is typically an irregular surface and cannot be described by mathematical expressions16. Digital terrain modeling is a numerical modeling process aimed at terrain topography, with its typical result being a DEM17, which is organized in a digital format according to a particular structure, representing the spatial distribution of actual terrain features as a 3D vector finite sequence model. It also serves as a digital description of the terrain’s shape, size, and undulations. The functional form of the DEM for the terrain in region D can be described as Eq. (1).18 where (Xi, Yi) ∈ D is the plane coordinate corresponding to the elevation point i; Zi is the elevation corresponding to (Xi, Yi); n is the number of elevation points. The construction of DEM can utilize various data sources, including stratum boundary line data, contour data, borehole data, section data, survey point data, and LiDAR measurement data, making it a commonly used data structure in open-pit coal mine geological modeling. The DEM representation model is divided into regular grid, contour, and TIN models, with TIN being the best and most commonly used method for detailed terrain modeling. By employing triangulation techniques for simplification and refinement (Fig. 4), the size and number of triangular facets can be controlled based on the complexity of the geological surface morphology, effectively eliminating data redundancy while maintaining a high level of fitting accuracy. Furthermore, TIN exhibits a good approximation effect for irregular boundaries such as faults, fractures, and other complex geological conditions. Under the premise of ensuring data integrity and correctness, it can generate high-quality DEMs, which is crucial for establishing fined geological models of slopes. Simplification (left) and refinement (right) of TIN. The edge collapse (A) and node collapse (B) operations coarsen the triangular mesh. Triangle subdivision (C) refines the mesh. Overview of B-Rep half-edge structure Traditional B-Rep records the geometric and topological information of 3D objects in hierarchical layers, establishing relationships between layers. Specific operations often involve a consistent sequence of listing the vertices on each face, which can be time-consuming for complex computations. To address this issue, Baumgart19 proposed the B-Rep based on the winged-edge data structure, which describes the topological relationships between an edge and its adjacent two vertices, four adjacent edges, and two adjacent faces. However, in the winged-edge data structure, two half-edges are adjacent to each edge, making it difficult to determine the current edge belonging to the loop and face due to the lack of edge orientation. Therefore, Roghothama and Shapiro20 improved and refined the traditional winged-edge structure by dividing each edge into two half-edges with opposite directions, each half-edge belonging to the edge loop of the two adjacent faces. Figure 5 illustrates its structure and corresponding storage format. The solid entity model data structure of the half-edge structure consists of five levels: solid, face, loop, half-edge, and vertex. It facilitates the retrieval of relevant information about the model components, making it easier to modify and manipulate local aspects of the model without altering the overall data structure. Objects with the same topological structure can be described using a unified data structure for different component surfaces, facilitating later-stage model restructuring and solidification, rapid verification of geometric topological relationships, non-geometric information integration, and model functionality expansion. Half-edge data structure and storage format. In constructing the 3D slope geological model of open-pit coal mines, the boundary surfaces constituting geometric entities such as terrain, strata, and faults are mainly achieved by clipping the TIN. The boundary of the B-Rep structure can be any edge and freeform surface, capable of defining entities through its boundary representation, providing effective solid descriptions for geometric objects. Applying the half-edge B-Rep data structure can construct a slope geological model under complex conditions through Euler operations based on the essential elements such as vertices, edges, and surfaces that make up the slope model (Fig. 6). Composition structure of B-Rep model. Overall modeling process In constructing a 3D geological model of open-pit mine slopes, it is necessary to first label the borehole and geological section data, divide and arrange the stratigraphic sequence, delineate the boundaries of geological interfaces, and then inspect and analyze the data to remove points with apparent errors. Due to the typically sparse nature of borehole data, appropriate spatial interpolation methods need to be introduced to interpolate the borehole data of each stratum, thereby establishing geological interface DEMs that better conform to actual geological conditions. Based on this foundation, fine modeling of the geological structure and distribution of faults in open-pit mines is conducted to generate a geological ground state model of the open-pit mine slopes. Utilizing mining design, geological reality, and UAV aerial survey data during the production process of open-pit mines, combined with dynamic updating methods of geological models, the integrated dynamic updating and reconstruction of the geological ground state model both above and below ground is completed. Finally, a refined 3D geological model of open-pit mines with faulted slopes is achieved (Fig. 7). Modeling process of 3D slope geological model. Multimodal data processing Source and characteristic analysis of geological data The 3D slope geological modeling dataset is typically diverse, sparse, and uneven, exhibiting multi-source heterogeneity. Based on different survey methods, the primary sources of geological data include geological survey data, remote sensing data, geophysical survey data, and exploration data. Among these, geological data obtained through geological exploration, especially borehole data, are the most precise. This data can accurately depict the spatial geographic distribution of lithological strata at borehole locations. Therefore, in 3D geological modeling, borehole data can serve as control data, while other geological data can serve as correction data, validating and correcting the 3D models established based on borehole data. Due to the diversity of data types and significant differences in accuracy, scale, and reference standards among data from different sources, the dataset usually exhibits characteristics of randomness and dynamics. Randomness: The complexity of the geological environment in open-pit mines, along with factors such as observation conditions, measurement errors, and human intervention, leads to much randomness of geological features. Randomness: The complexity of the geological environment in open-pit mines, along with factors such as observation conditions, measurement errors, and human intervention, leads to much randomness of geological features. Dynamic: In most cases, geological structures cannot be objectively, accurately, and in real-time reflected, resulting in geological data appearing “gray”. However, with the advancement of geological re-surveying and production activities in open-pit mines, new sampling point data continues to emerge, gradually revealing the stratum of open-pit mines. As a result, the geological conditions of the slopes become clearer, and the accuracy of the data improves gradually. Dynamic: In most cases, geological structures cannot be objectively, accurately, and in real-time reflected, resulting in geological data appearing “gray”. However, with the advancement of geological re-surveying and production activities in open-pit mines, new sampling point data continues to emerge, gradually revealing the stratum of open-pit mines. As a result, the geological conditions of the slopes become clearer, and the accuracy of the data improves gradually. Given the above characteristics, the interpretation of geological data, interpolation of sampling points, and dynamic updating of geological models become particularly important. Subsequent chapters will further analyze and apply interpolation of sampling points and model updating methods. Sequence stratigraphic division In 3D geological modeling, the most prominent feature of geological bodies is their multi-layered nature, where the distribution characteristics of these layers determine the geometric structure of the geological model. When conducting sequence stratigraphic division, it is necessary to determine the division criteria based on the geological phenomena and evolutionary characteristics within the study area, combined with appropriate geological interpretation method theories. The sequence stratigraphic division is generally carried out by experienced geological engineers who integrate borehole data, well logging data, section data, and seismic data. They determine the spatial distribution of various strata in the geological body through geological interpretation and rational inference. The specific method involves analyzing the geological evolution history obtained from comprehensive borehole core columnar diagrams to establish a sequence of stratigraphic layers. Then, the level set theory is applied to topological operations, forming a 3D stratigraphic division strictly based on the geological evolution history21, as depicted in Fig. 8. Sequence stratigraphic division. 3D stratigraphic spatial interpolation Borehole data is the most reliable source for 3D geological modeling. However, due to economic considerations, the number of boreholes in the study area is generally very sparse and limited, making it difficult to describe the geological conditions accurately. Therefore, spatial interpolation of borehole data is necessary to solve this problem. Spatial interpolation methods can be categorized into geometric methods, statistical methods, spatial statistical methods, functional methods, stochastic simulation methods, physical model simulation methods, and integrated methods22, each with advantages, disadvantages, and applicable conditions23. When conducting actual spatial data interpolation, it is necessary to analyze the relationships and regularities among the original data while also considering factors such as the uncertainty of the interpolation results, the smoothness of the surface, and the interpolation speed. Appropriate interpolation methods should be chosen by combining the principles of geography and relevant geological conditions. Commonly used 3D spatial interpolation methods in open-pit mine geological modeling include the IDW, Kriging, trend surface, weighted least squares fitting methods, etc. The IDW method belongs to a statistical method in geosciences. This method utilizes the distance between the measured points and the known values as weighting factors to infer the values of unknown points based on the data values of known points24, which can be expressed as follows: where Z(x) is the estimated value of the interpolation point, Z(xi) is the value of the i'th interpolation point, n is the total number of participating interpolation points, Wi is the weight value of the i'th interpolation point, and the weight Wi decreases with the increase of distance, which can be written as follows: where d(x, xi) is the distance from the i'th interpolation point xi to the target point x, α is the distance exponent value. If the value of α is too large, the weight of the interpolation points closer to the interpolation point will be increased, making the interpolation value close to the value of the nearest interpolation point. Conversely, the interpolation value will be closer to the average value of the surrounding interpolation points. The criterion for selecting α is to minimize the mean absolute error. For coal seam deposits, it is generally set to 2. The IDW method is susceptible to the influence of data point sets. In practical applications, it is necessary to apply limiting conditions to the IDW method according to different situations. For example, combining the search radius and search direction to search and evaluate sample points can improve the accuracy and rationality of spatial scattered data interpolation. Considering the IDW method’s excellent versatility, it is applicable even in missing strata and highly uneven distribution of boreholes, with relatively small interpolation errors. Therefore, we chosen the IDW as the fundamental interpolation method for spatial borehole sampling data. TIN algorithm construction The construction of the TIN algorithm is the basis for generating DEM. Currently, the most commonly used method for TIN generation involves Delaunay triangulation under the criteria of empty circumcircle and maximum minimum angle. Due to its uniqueness and favorable triangular properties, it is considered the most suitable for surface approximation25. The methods for constructing TIN can be classified into three categories: divide-and-conquer algorithm, triangle mesh growth algorithm, and point-by-point insertion algorithm. Among them, the point-by-point insertion method is a dynamic triangulation construction algorithm. It has a simple implementation and can effectively overcome the selection problem of borehole data. Moreover, it meets the demands of large-scale open-pit mine model data sources and subsequent dynamic update reconstruction of stratigraphic interfaces. Therefore, the point-by-point insertion method is adopted for generating DEMs of various strata in this 3D geological modeling process. Open-pit mine terrains often exhibit complex step-like distribution characteristics. To construct high-precision DEM terrain models, terrain feature lines are typically used as constraint edges for mesh construction. Otherwise, the generated DEM may suffer from significant distortions, leading to discrepancies between the DEM and the actual terrain, such as the flattening of benches. Constrained Delaunay triangulation (CDT) is established based on Delaunay triangulation research. By embedding constraint lines to form a constrained triangulated network, CDT considers the constraints of features like benches, thus providing a more realistic simulation of step-like morphology in open-pit mines. Building upon the establishment of non-CDT using the point-by-point insertion method, we utilize the multiple diagonal exchanging algorithm26 from the two-step method27 to forcibly embed constraint edges into the pre-constructed triangulated network to construct CDT. The flow of the TIN algorithm is demonstrated in Fig. 9. TIN algorithm flow. This algorithm enables the generation of unconstrained and constrained triangulated networks while facilitating dynamic updates of the current TIN within a local scope. In scenarios where refinement or updating of local borehole point cloud data is required to obtain a more detailed 3D geological model due to practical needs, interactive operations can be integrated into the algorithm. By acquiring the encrypted or updated area range, the algorithm can reconstruct the DEM of the corresponding stratum within that range, generating a refined or updated slope geological model within the local scope. Fine construction method of fault Faults are a typical geological structure in the Earth’s crust, and fault models describe the geometric shapes of faults and their mutual contact relationships while also providing constraints for generating stratigraphic surfaces28. The ore-forming process in open-pit coal mines is usually accompanied by complex geological structural movements, resulting in frequent fault fracture zones and structural surfaces near ore veins. The occurrence of faults disrupts the integrity of slopes, forming spatial structural surfaces that divide the slopes into two parts. Slope sliding often occurs on these structural surfaces, significantly compromising the stability of the slopes29,30. Therefore, the accuracy and fine of fault models directly affect the reliability of slope geological modeling and stability analysis. Based on different survey methods, data accuracy, and types, fault data can generally be divided into fault point, fault polygon, and fault stick data. Fault point data are obtained from borehole data and actual exposures, which are relatively sparse and make it challenging to construct complex fault morphologies. Fault polygon data are inferred from 3D seismic or geological observations, often with overlapping areas, posing limitations in fault modeling. Fault sticks represent the parallel parallel section of faults in the depth domain. Extracting fault sticks from each geological section map yields a 3D wireframe model of faults, with data interrelated and exhibiting high accuracy and reliability. When modeling faults, fault stick data can be chosen as the fundamental control data for fault surface reconstruction, and other data can serve as corrective data for verifying and refining the established fault models. However, due to economic constraints, original exploration line sections in open-pit mines often have large spacing, resulting in sparsely distributed fault sticks. Moreover, fault morphologies are often complex, and reconstructing fault surface morphologies based on traditional outline sticks methods is too rough to effectively capture the curved details of fault points, significantly affecting the accuracy and practicality of 3D geological models. Morphing transformation can simultaneously consider the shape and other features of the objects involved in the transformation. It was initially applied in computer image processing31, allowing a smooth transition from the original to the target image. The steps for constructing a refined fault surface based on morphing transformation are as follows: Feature point matching. Feature point matching. Implementing feature point matching is critical in morphing for a smooth transition. For fault stick data, feature points include the intersection points of the faults with the tops and bottoms of various coal seams, as well as other control points that ensure the integrity of the fault morphology. Since the number of control points may vary between adjacent fault sticks, it is crucial to match them to ensure each has corresponding feature points accurately. As illustrated in Fig. 10, the relationships between feature points can be categorized into four cases: one-to-one (a), many-to-one (b), many-to-zero (c), and many-to-many (with varying quantities) (d). For cases (a) and (b), feature point matching can be directly implemented. However, for cases (c) and (d), feature point matching must be conducted based on the curvature characteristics of the curves. Corresponding relationship of feature points. To simplify the matching process, additional points can be added to ensure each fault stick has an equal number of control points Pij. If fault stick Gi has a maximum of k points and fault stick Gi+1 has m points (where m < k), then Gi+1 requires adding k-m points. To ensure that the nodes are evenly distributed along the segments, points can be added at the midpoints of the longer segments in G. By performing k-m binary insertions based on segment lengths, the distances can be refined, allowing for a one-to-one correspondence of Pij across each fault stick. (2)Morphing interpolation transform. Morphing interpolation transform. Utilizing the established feature point matching relationships, morphing interpolation transformations are performed to create transition curves between adjacent fault sticks. The control point function for the transition curves can be described as follows: where L(i,t) is the interpolated point corresponding to the starting fault point S(i) and the ending fault point A(i); t is the displacement distance value (t ∈ (0,1)), indicating the degree of separation between the interpolation point on the transition curve and S(i) and A(i). When t = 0, the Morphing transformation interpolation result is S(i); when t = 1, the interpolation result is A(i). Finally, the fault surface is generated by CDT, and the modeling process is shown in Fig. 11. The fault TIN strictly adheres to the actual fault sticks and is constrained by transition curves. Consequently, the resulting fault model exhibits higher accuracy, with the fault morphology well controlled. However, the control points of the fault sticks can significantly affect morphing transformations, making it challenging to maintain the fault model’s accuracy, especially in cases of particularly sparse data. Construction process of the fault model. Geological ground state model construction of slope The geological ground state model refers to the initial 3D geological model of an open-pit mine before any mining activities occur, generated by applying external boundary constraints to the surface and various strata. During the modeling process, the cutting action of faults results in discontinuities in the spatial interfaces of the strata, leading to two types of contact relationships based on the objects involved: fault-to-fault and fault-to-stratum. The prerequisite for the correct operation of meshing algorithms is that no two surface patches intersect; if two surfaces intersect, the intersecting edge must be part of the input surface32. Therefore, corresponding intersection-cutting methods should be employed based on the different contact relationships to ensure the strict topological consistency of the interfaces in the generated geological model. After completing each stratum’s modeling and intersection-cutting constraint processing within the slope modeling area, each stratum model’s maximum internal plane range and height range in 3D space are determined. Based on these ranges, a closed space cubic domain model is constructed, and external boundary constraints are applied to the model to complete the geological ground state model construction. Processing of fault-fault intersection-cutting relationship In complex fault systems, there are numerous faults with intricate structures. Multiple faults intersect and cut each other, forming a network of faults. For intersecting strata, it is essential to analyze the trends and orientations of fault directions thoroughly. Based on this analysis, it is necessary to employ rational extrapolation methods to extend fault surfaces to establish correct intersection and combination relationships between faults. For fault stick data, the extension direction of the fault primarily lies in the strike direction. Based on the established matching relationship of characteristic points between fault sticks, a simple, effective, and easily implementable Lagrange interpolation method can be used for trend extrapolation of fault sticks. Let (x0,y0,z0),(x1,y1,z1),…,(xn,yn,zn) be n + 1 sets of coordinate points with different abscissa values. The curve can be defined as: where Ln(x) determines the extension direction angle of the curve, while Qn(x) determines the trend of elevation changes along the curve. Since higher-order interpolation may lead to Runge’s phenomenon, causing the fitted curve to deviate from the expected behavior, it is advisable to use low-order Lagrange interpolation methods whenever possible. Meanwhile, the extension portion should consider the fault surface’s overall trend and local variations. Typically, it is recommended to establish the curve using 2 or 3 points corresponding to the extension endpoint for extrapolation. The extension distance should not be excessively long; the greater the distance, the larger the error. It is advisable to extend slightly beyond the main fault. In order to ensure the topological consistency between fault grids, intersectional-cutting constraint processing must be performed on fault surfaces. The specific method involves determining intersection lines between two fault surfaces, adding these intersection lines to the respective fault surfaces according to the CDT criteria, and using these intersection lines as paths to trim secondary fault segments extending beyond the main fault, thus achieving intersection processing of the faults (Fig. 12). Processing flow of fault surfaces intersection-cutting relationship. Processing of fault-stratum intersection-cutting relationship In constructing a geological interface model, the stratum surface transforms from a continuous surface to a discontinuous one due to fault cutting. Hence, it is imperative to account for the cutting effect of faults when establishing the stratum surface model and to determine the specific morphology of the intersection line between the stratum and faults. The intersection lines are often inferred by geological personnel based on geological knowledge, which can lead to frequent logical errors. Additionally, the extrapolated inference sections often need a more rigorous theoretical basis. This deficiency results in constructed fault surfaces that do not align well with the stratum interfaces, consequently generating gaps. Such gaps undermine the effectiveness of macroscopic contact of geological interfaces in slope geological models, rendering it impossible to guarantee topological consistency at intersecting positions. In response to the issues above, it is proposed to address them by selecting stratum triangles near the faults and extending them according to specific rules to intersect with them. This approach aims to obtain a method for deriving logically consistent intersection lines between stratum and fault surfaces, thereby achieving the seamless integration of stratum and fault models. The basic principle of the method is shown in Fig. 13, where T1 to T7 are edge stratum triangles near the fault, P1 to P4 are points to be extended on the outer boundary, S1 to S5 are points on the boundary triangles located internally, and E1 to E4 are extended points obtained through each point to be extended. Processing of fault-stratum intersection-cutting relationship. Assuming the current point to be extended is P2, the extension direction at P2 should point toward the exterior boundary. The calculation formula for the extended point E2 is as follows: where is the average trend vector of the fault; is the average outward normal vector of all triangles with the current point as the vertex; Lstep is the extension step length. Equation (7) satisfies the right-hand rule, ensuring that the extended point E2 is located outside the stratum surface. The use of this formula generates triangles gradually converging towards the fault surface. On the other hand, Eq. (8) facilitates the natural extension of the surface during the extension process, considering the undulating changes of the stratum at the fault. After generating new triangles through the triangle mesh algorithm for the points to be extended and their extensions, the intersection line with the fault surface is calculated; subsequently, this intersection line is incorporated into the fault surface and stratum interface according to the CDT criterion. The excess stratum mesh beyond the fault surface is trimmed along the intersection lines. Following the sequence from top to bottom of the stratum interface, intersection-cutting constraints are sequentially applied between the fault surfaces and stratum surfaces to obtain the updated fault surface and stratum interfaces. Dynamic update method of slope geological model With the advancement of open-pit mining operations, geological information is continually revealed, leading to the accumulation and updating of geodetic data in existing 3D geological models. This influx of data enhances the accuracy of these models in representing underground structures. Therefore, the dynamic updating process of slope geological models should involve continuous iteration and refinement, gradually transitioning the geological model from ambiguous to transparent33. Model updates primarily focus on surface and subsurface aspects of open-pit mines. Dynamic update of surface model Surface models are generally divided into open-pit design surface models and UAV survey acceptance surface models, depending on their data sources and purposes. The former is generated from mining engineering plans designed by mining engineers to evaluate slope stability during mine construction. The latter is often generated from UAV LiDAR point cloud data to assess slope stability during mining. Both models undergo frequent updates to meet the requirements of production planning and survey acceptance in open-pit mines. Therefore, multiple-period surface models can form a continuously updated 3D geological model of slopes (Fig. 14), thus catering to various demands for slope stability analysis during the production process of open-pit mines. Dynamic update process of 3D surface models. Dynamic update of stratigraphic interface During the production process of open-pit coal mines, the geological strata beneath the surface gradually become exposed, revealing these strata’s actual morphology and spatial position information. In order to meet the demand for dynamic updating of stratum surfaces in the slope geological ground state model, we propose a method for locally updating and correcting slope geological models based on actual stratigraphic point data. The specific steps are as follows: Vertical projection of the polyline formed by connecting sample points in the Z-direction to generate projection intersections. Subsequently, an inclusion analysis is performed on the triangles intersected by these lines to determine their corresponding triangles. Vertical projection of the polyline formed by connecting sample points in the Z-direction to generate projection intersections. Subsequently, an inclusion analysis is performed on the triangles intersected by these lines to determine their corresponding triangles. Utilize a layered search algorithm centered on the identified triangle to execute a tiered search. The number of search layers can be determined based on the ratio of the modified influence range to the spatial interpolation distance. The modified range should comprehensively consider various factors, including the spatial distribution of open-pit mining modeling data, the stage advancement distance, and the data collection update cycle. Utilize a layered search algorithm centered on the identified triangle to execute a tiered search. The number of search layers can be determined based on the ratio of the modified influence range to the spatial interpolation distance. The modified range should comprehensively consider various factors, including the spatial distribution of open-pit mining modeling data, the stage advancement distance, and the data collection update cycle. Remove triangles within the correction scope, using the TIN open edges of the correction scope as a constraint. Combine engineering geological reality and borehole sample points for re-interpolation processing and generate a triangulation network. Remove triangles within the correction scope, using the TIN open edges of the correction scope as a constraint. Combine engineering geological reality and borehole sample points for re-interpolation processing and generate a triangulation network. Topological consistency processing is performed at the junctions of the new and old meshes to generate a locally updated 3D geological model. Topological consistency processing is performed at the junctions of the new and old meshes to generate a locally updated 3D geological model. The dynamic updating process of the model is shown in Fig. 15, with a spatial interpolation distance of 50 m and one layer of search. This method demonstrates a significant corrective effect on the geological model through field validation. Dynamic updating process of geological interface. Case study To validate the effectiveness of the slope model modeling and updating method, we use the northern side slope of an open-pit coal mine in Inner Mongolia, China, as a case study. The research area is delimited to include slopes containing the Y-shaped intersection fault and faults of the open-pit coal mine. The modeling steps are as follows: (1) Utilize engineering geological information within the area to complete stratigraphic division and establish a geological geometric structure database. (2) Conduct spatial interpolation of the stratigraphy using the IDW method. Generate stratigraphic interfaces and fault surface meshes through the TIN algorithm and perform extrapolation (Fig. 16①). Here, Su is the initial surface of the open-pit mine, Fx1 and Fx2 are the fault surfaces, and the bottom interfaces of the stratigraphic layers from top to bottom are as follows: Quaternary system (Qdown), Neogene system (Ndown), sandstone (Sdown), upper mudstone (M1down), coal seam (Cdown), and lower mudstone (M2down). (3) Utilize the model intersection-cutting processing method and rules (Fig. 16②) to construct the geological ground state model M of the slope (Fig. 16③). (4) Update M using the current state surface SS (Fig. 16④), generating a closed topologically consistent 3D slope geological model (Fig. 16⑤). In practical applications, the above model can serve as the basic model. After adjusting and optimizing the mesh and tetrahedral subdivision, it can be imported into FLAC3D 6.00 through format conversion. The physical and mechanical parameters of the rock layers are assigned to the tetrahedral mesh model based on the rock mass mechanical parameter table, assigning cohesion, friction angle, capacity, and other mechanical parameters to the corresponding stratum models (Fig. 16⑥). As open-pit mining progresses, the surface conditions and geological reality data are continuously updated. The Geological ground state model is gradually iterated and corrected, providing a reliable foundation for slope stability analysis and optimization of slope angles in open-pit mines. Construction of the 3D geological model of the slope. Conclusions The following conclusions and comments can be made from the present study: The essential elements and modeling rules of slope geologic modeling are explored. A 3D data structure for geologic modeling of complex slopes containing faults in open-pit coal mines is proposed, which utilizes the DEM to construct 3D structural interfaces for slope engineering, employing the half-edge B-Rep structure to organize the spatial topological relationships of the DEM. The geometrical description of the geological structure, ground, and strata composing the 3D model of the complex slope is realized. The essential elements and modeling rules of slope geologic modeling are explored. A 3D data structure for geologic modeling of complex slopes containing faults in open-pit coal mines is proposed, which utilizes the DEM to construct 3D structural interfaces for slope engineering, employing the half-edge B-Rep structure to organize the spatial topological relationships of the DEM. The geometrical description of the geological structure, ground, and strata composing the 3D model of the complex slope is realized. After analyzing the source and characteristics of the open-pit mine’s slope data, the sequence stratigraphic division and spatial interpolation methods of slope geological modeling are studied, and a TIN dynamic growth construction algorithm suitable for building geological models of open-pit coal mine slopes is proposed. A refinement reconstruction method based on Morphing transformation was proposed to address the rough and inaccurate morphology of fault surfaces constructed from traditional sparse fault stick data, achieving smooth construction of fault surfaces under sparse data conditions. After analyzing the source and characteristics of the open-pit mine’s slope data, the sequence stratigraphic division and spatial interpolation methods of slope geological modeling are studied, and a TIN dynamic growth construction algorithm suitable for building geological models of open-pit coal mine slopes is proposed. A refinement reconstruction method based on Morphing transformation was proposed to address the rough and inaccurate morphology of fault surfaces constructed from traditional sparse fault stick data, achieving smooth construction of fault surfaces under sparse data conditions. Methods for handling the intersection-cutting of faults with each other and strata in 3D slope geological modeling are presented and used Lagrange interpolation to extrapolate the trend of fault line node curves, achieving complex intersection-cutting handling between faults and constructing fault network models. By extending the initial stratum surface, we obtained the constrained intersection lines between stratum and fault surfaces, addressing the intersection-cutting relationship between stratum and fault surfaces and ensuring the seamless integration of geological models, thus maintaining the topological consistency of the geological model. Methods for handling the intersection-cutting of faults with each other and strata in 3D slope geological modeling are presented and used Lagrange interpolation to extrapolate the trend of fault line node curves, achieving complex intersection-cutting handling between faults and constructing fault network models. By extending the initial stratum surface, we obtained the constrained intersection lines between stratum and fault surfaces, addressing the intersection-cutting relationship between stratum and fault surfaces and ensuring the seamless integration of geological models, thus maintaining the topological consistency of the geological model. With the transparency of strata geological information during the open-pit mining process, a method and process for dynamically updating the base state model of slope geology using surface geological correction data are proposed, allowing for the updating and reconstruction of the 3D geological model. The method enhances the transparency and practicality of the refined model. However, geological model updates in areas where faults and adjacent strata intersect remain unresolved. Furthermore, additional research is needed to achieve the integration and visualization of space-air-ground slope monitoring data while continuously optimizing model construction algorithms for greater simplicity and efficiency. With the transparency of strata geological information during the open-pit mining process, a method and process for dynamically updating the base state model of slope geology using surface geological correction data are proposed, allowing for the updating and reconstruction of the 3D geological model. The method enhances the transparency and practicality of the refined model. However, geological model updates in areas where faults and adjacent strata intersect remain unresolved. Furthermore, additional research is needed to achieve the integration and visualization of space-air-ground slope monitoring data while continuously optimizing model construction algorithms for greater simplicity and efficiency. Acknowledgements This research was supported by the National Natural Science Foundation of China (52374123), the Youth Fund of the National Natural Science Foundation of China (52204158), and the Basic Research Project of Liaoning Provincial Education Department (LJ212410147019). Author contributions J.Y. wrote the main manuscript text. G.L. wrote the original model. All authors reviewed the manuscript. Data availability The data used and analysed during the current study available from the corresponding author on reasonable request. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The data used and analysed during the current study available from the corresponding author on reasonable request.