MENU
2023 lmpact Factor: 4.6 (Q1)
2023 CiteScore：7.8 (Q1)
2023 CiteScore：7.8 (Q1)
Citation:
Ma Zhao-yang, Zhang Sen, Li Jie, Qiao Ya-kai, Sun Hua-feng, Bouchaala Fateh. 2025. A new integrated 3D modeling method based on multisource heterogeneous geological data and its application. China Geology. doi: 10.31035/cg2024063.
|
As new strategic initiatives for prospecting breakthroughs advance, there is an increasingly urgent need to explore the deeper and peripheral areas of mineralization belts. Specifically, the focus of exploration has shifted from outcrops, shallow layers, and easily identifiable ore bodies to concealed, deep, and challenging targets. Given the rising complexity of exploration, it has become imperative to adopt innovative theories, technologies, and methods to improve the efficiency of mineralization prediction and exploration. A recently developed 3D geological modeling method that integrates surface and remote information has emerged as a crucial tool in mineralization prediction (Yu RQ et al., 2013). Combining remote sensing images, digital elevation models (DEMs), and 3D geological modeling effectively enhances the visualization of geobodies. The key aim of 3D geological modeling is to provide an accurate and objective representation of geological conditions by integrating multisource data. Currently, popular 3D modeling software packages such as Petrel, Surpac, and Micromine suffer from a common limitation: they fail to simultaneously load and process multiple data types (Zhou F et al., 2022). In recent years, many researchers have proposed and improved various 3D geological models based on extensive studies and practical applications. Typical examples include a borehole-based stratigraphic boundary model (He HJ et al., 2002), an object-oriented vector-raster integrated 3D model for ore bodies (Liu YJ et al., 2012), a profile-based method using tunnel-stratigraphy intersections (Zhang ZH et al., 2014), and a virtual generalized triangular prism (GTP) model (Li CL et al., 2013). Although these models have advanced the modeling of complex geobodies, they remain limited by dependence on single data sources and low precision.
Using big data and AI, this study proposes a novel 3D modeling method based on multisource heterogeneous geological data processing and high-precision constraints. To validate its practicality, this method is employed to conduct the 3D geological modeling of the X area within the Sichuan Basin using remote sensing images, geological maps, geophysical data, borehole data, and XRF spectroscopy data. Modeling and visualizing the X area reveals a new mineralization belt, confirming the feasibility and effectiveness of the new approach.
Information technology (IT) is rapidly transforming production, lifestyles, and thought while also advancing Earth system science and modernizing geological practices. Since the establishment of the China Geological Survey, the digitalization of geological surveys has been one of its key priorities (Li CL et al., 2015). Recently, traditional methods for representing geological information, such as maps and cross sections, have struggled to meet the demands of big data and digital transformation. Research on 3D geological modeling and visualization has attracted wide attention across multiple disciplines, supporting the ‘Glass Earth’ strategy and the integration of multisource heterogeneous data in geological big data in several countries (Zhang Y et al., 2022). Moreover, 3D visualization has increasingly become a significant component of the digitalization of geological surveys. New 3D geological modeling methods are systematic approaches that integrate technologies including multisource heterogeneous data processing, holistic model databases, seamless local modeling, remote sensing, and the convergence of big data with AI.
Geological big data technology is a technical system that integrates cloud computing, information and communication technology, and internet platforms to collect, manage, analyze, and apply geological data. By employing advanced data processing and analysis techniques, this technology reveals potential value in geological data, thus providing robust support for geological studies, resource exploration, early warning of geologic hazards, and environmental protection. Its application to 3D modeling enhances data processing while also ensuring model integrity (Wang Y et al., 2021). However, the data employed are characterized by multiple sources and heterogeneity. Specifically, geological phenomena are represented using various data types, including remote sensing, borehole, cross-sectional, and geophysical exploration data. Furthermore, each of the data types potentially has multiple sources. As a result, even for the same geological phenomena in a given area, it is challenging to ensure the consistency of multisource heterogeneous geological data. The resulting uncertainties in data quality necessitate thorough processing to construct accurate 3D geological models.
The fusion of multisource remote sensing data is a technology that comprehensively processes and analyzes these data, particularly facilitating the integration of these data with other multisource information in geoscience research (Ling H et al., 2005). Multisource heterogeneous data processing plays a fundamental role in geological modeling. Regardless of the data source, the key to the processing is to convert heterogeneous data into isomorphic vector data (points and lines), thereby providing foundational data on surface features such as ground, fault planes, and rock mass surfaces.
During geological modeling, when data conflicts occur, the multisource heterogeneous data processing technology determines the priority of conflicting data, with higher-accuracy data corresponding to higher priority. In other words, the usage of the original data is positively correlated with their accuracy.
Large-scale, high-precision holistic modeling contributes to the establishment of a continuous 3D geological model that meets the accuracy demands of various geological research fields. This approach depends primarily on model database and data expansion technologies. Model database technology facilitates the integration of different strata in a large area, ensuring that local models achieve accuracy comparable to traditional block models. In contrast, data expansion technology allows for the splicing and fusion of boundary faults and strata, thus improving fault delineation. Using complex structures, large-scale and high-precision holistic modeling ensures consistent model information, high-precision fusion across areas, and effective inheritance of existing modeling achievements. Additionally, holistic modeling enables unified storage of data and models while establishing a unified management and communication platform. This technology is suitable for exploration and exploitation in the study or mining areas that offer extensive, diverse geological data. The model constructed using this technology can accurately represent all geological structures in the study area. Besides, the holistic model can serve as the basis for fine-scale geological management. Lastly, large-scale, high-precision holistic models can provide accurate geological descriptions, rapid computations, and dynamic updates. Fig. 1 shows the working principle of large-scale, high-precision holistic modeling. Initially, raw data are stored in a database. Then, a holistic model is constructed using holistic model database technology, seamless local modeling, and a collaborative parallel working pattern. Third, cluster visualization technology is used to describe the entire model, supporting the development of various professional applications (Ma ZY et al., 2022).
The holistic model database technology employs a chip storage mode to generate and store models as field data, transforming the traditional storage file mode of models. This represents the core of holistic model databases. During model storage, users specify and separate model cells in the external display first, and the model cells separated are then saved to the database file. Specifically, each cell is stored in a corresponding table, whose information is recorded as fields. When a local model is downloaded, the system identifies the locked cells of the model. Therefore, an established local model is derived into multiple slice or cell models after being processed. In the database, the information of these models is recorded as fields in corresponding cell tables. The call process of models is similar to the storage process. Specifically, the locked cell models are initially extracted from the database and then processed to form a complete model (Fig. 2).
The splicing of areas with few splices can be achieved using traditional modeling methods. However, excessive and overly large splices will lead to an overlarge working area to be processed by conventional software. In this case, model database technology should be adopted. This technology enables seamless splicing of local models using local splicing and extended grids, thus achieving the splicing of multiple or large working areas. The seamless splicing technique for local models involves loading all the data of an overlapping area into the database to achieve a unified understanding of structures. Then, the model of the overlapping area is updated, during which the updated model and its surrounding models can be seamlessly spliced using extended grids. In this manner, seamless splicing and fusion across multiple working areas can be achieved.
In local modeling, it is necessary to select proper splicing methods for various splicing sites. For the splicing sites involving faults, splicing can be achieved as follows: (1) When faults at the splicing sites remain unchanged, seamless splicing can be performed automatically; (2) When faults at the splicing sites change without affecting adjacent blocks, seamless splicing can also be executed automatically; (3) When faults at the splicing sites change and adjacent blocks are affected, an additional block at the fault should be downloaded and remodeled to achieve seamless splicing. The stratigraphic models in the splicing areas can be established only when the fault and stratigraphic data are consistent. This allows for a unified understanding of boundary faults across adjacent blocks, effective integration of modeling results, and avoidance of repetitive modeling (Ma ZY et al., 2022). Fig. 3 shows the identification of boundary faults in an overlapping area.
During holistic modeling, the collaborative parallel working pattern can be applied to divide the working area into multiple segments, allowing the parallel construction of a holistic model by multiple people. This pattern overcomes the limitations of the conventional single-person method in achieving regional modeling.
The cluster visualization system is a platform focusing on cluster graphics processing. It can provide the visualization of both the large-scale holistic model and local models of any size and location while also allowing the generation of the geological plans and sections of any location.
Remote sensing technology employs various types of sensors located at different distances to obtain the Earth’s information. Since the acquisition of the first remote sensing image, this technology has been applied to land use, hydrology, geology, meteorology, climate studies, and vegetation studies (Horning N et al., 2008; Slonecker ET et al., 2001). Its evolution has significantly broadened the horizon of field-related technologies, establishing it as an indispensable tool for geological studies and 3D modeling. The development of sensor parameters has significantly improved the observation scale, resolution, and identification accuracy. Remote sensing geology has transitioned from macroscopic to microscopic detection and from qualitative interpretation to quantitative inversion, resulting in elevated applications (Wang RS, 2011). Its primary research directions include remote sensing image processing, multi-spectral alteration anomaly extraction, and hyperspectral mineral identification and mapping. Additionally, it also involves surface deformation survey and monitoring based on interferometric synthetic aperture radar (InSAR), remote sensing-based ore prospecting models, and remote sensing-based geologic hazard investigation and monitoring. In this context, the research results of remote sensing technology are essential for predicting the structural morphology, lithology, and attribute distribution of 3D geological models (Du XF, 2018). A major advantage of this technology is its ability to identify and extract the spatial extent and elemental characteristics of built-up areas, thus supporting planning and resource management, risk assessment, and the early warning of disasters (Ge W et al., 2018; Bhaskaran S et al., 2010).
Given the close relationship between fault formation and rock mass interactions, the fault structures of adjacent rock masses can be analyzed and compared based on their spatial distributions. For instance, using the Enhanced Thematic Mapper Plus (ETM+) remote sensing images as a data source, information about linear structures can be extracted from the remote sensing images of rock masses. By comprehensively analyzing isodensity maps, trend surfaces, and stress field distributions, the genetic relationships of rock masses can be determined. Remote sensing image data are images that are acquired from various sensors and undergo error processing and radiometric correction. After constant advancement, modern remote sensing technology has become a significant tool for extracting information from images due to its macroscopic, dynamic, rapid, accurate, and comprehensive nature (Li XC, 2005). Satellite image products play a crucial role in map-assisted datasets (Stevens FR et al., 2015). Surface DEMs are primarily acquired using methods including field surveys, map digitization, and InSAR (Wasklewicz T et al., 2013; Zhang W et al., 2016). Moreover, with the rapid development of IT, many DEM datasets can be downloaded directly on the Internet (Rott H et al., 2012; Woodcock CE et al., 2008).
In the new 3D modeling method developed in this study, a major improvement in remote sensing image processing is the registration of remote sensing images with DEMs. This registration serves as the basis of 3D geomorphologic image mapping. For the superposition of remote sensing image data and DEMs, geometric registration is performed initially to transform the coordinate spaces into a unified projection coordinate system. In practice, a control point-based registration method is employed to match the control points and estimate the geometric transformation parameters for registration. Then, the registered remote sensing images are mapped as texture data onto the DEM-based 3D model surface, enabling realistic and objective landscape simulation. The spectral characteristics of remote sensing images can directly affect the visual quality of 3D geomorphologic models, while the accuracy of image-DEM overlapping can determine the accuracy of the 3D model reconstruction. Therefore, remote sensing image processing is crucial to improving the overall accuracy and fidelity of 3D models. Regarding 3D terrain visualization, with the remote sensing images of a certain area as textures, the DEM and the images are superimposed while ensuring coordinate registration among different datasets. As a result, data elements like text symbols, geographic features, and images of the area from various sources are transformed to share a unified coordinate system, contributing to the generation of the real-time 3D terrain of the area. The integration of remote sensing images with DEMs involves three key processes: the generation of the DEMs from remote sensing images, the correction of remote sensing images using the DEMs, and the registration between remote sensing images and DEMs (Han L, 2005).
Remote sensing-based geoscience analysis aims to provide more intuitive information, identify deeper spatial relationships, and enhance the accuracy of image classification and special feature extraction by integrating remote sensing data, geographic information, and geoscience knowledge. Developing spatial geodetic techniques is crucial for measuring surface displacement, thus increasing the number of detectable areas subjected to deformations (Fernández J et al., 2018; Huang Y et al., 2015; Yu M, 2017; Biggs J et al., 2014; Biggs J et al., 2018). In geology, remote sensing interpretation comprises four major aspects. (1) Interpretation of vertical faults: it involves assessing the changes in both their elevation differences and stratigraphic properties. For instance, NASA’s Landsat 7, equipped with ETM+ sensors, provides data with higher spatial resolution and spectral richness compared to earlier emissive and land cover data. Such satellite data can serve as a reference for validating various methods in the future (Ma XL et al., 2017; Jiang W et al., 2017). Therefore, Landsat—with its long time series and high spatial resolution—can serve as a valuable data source (Wang KL, 2019); (2) Interpretation of horizontal strike-slip faults: horizontal displacements produced by strike-slip faults have a wide range and long-time span. Features such as offset ridges, asymmetric valley slopes, and synchronous turning of water streams and terraces are key indicators for identifying horizontal strike-slip faults (Li JC, 2014); (3) Interpretation of multistage faults: multistage faults are complex fault structures formed by multistage tectonic movements, occurring as T- or Y-shaped structures; (4) Interpretation of strata: remote sensing images can vividly capture the structural traces and spatial framework of crustal movements, providing multi-level, multi-scale, and multisource structural landscapes for research on regional structures.
Urban development and construction lead to increasing use of underground spaces, altering geobody shapes. To accurately characterize geobodies and assess their interactions with man-made structures, the impacts of man-made structures on geobodies must be examined first. The fusion of geological models and DEM data aims to address the mismatch between high-precision DEM surfaces and low-precision geological model grids. This can be achieved by cutting the geological mesh using a high-precision DEM surface. As a result, a new, complex polyhedral mesh that integrates the DEM and geobody data can be formed.
Traditional large-scale 3D geological models are only applied to simple macroscopic visualization due to their low accuracy. In contrast, the new 3D geological modeling method developed in this study can satisfy diverse needs at both macroscopic and microscopic scales.
Holistic structural modeling technology is crucial to the geological suitability assessment of large-scale engineering projects. This technology enables the integration of multisource data under the foundational framework of a holistic structural model. As a comprehensive platform for collaboration among multidisciplinary professionals, holistic structural models meet assessment needs in both the scale and precision dimensions. The processing technology for multisource heterogeneous data provides an accurate geological model for assessment. Notably, the construction of attribute models supplies a wealth of spatial attribute data, supporting quantitative assessments.
Remote sensing-based image data mining technology is to discover the information hidden in remote sensing images using image analysis, pattern recognition, AI, geographic information systems, and spatial data mining theories. This technology is the application of image data mining to remote sensing. The analysis of geological big data mining has become a hot research topic, especially with the increasing maturity of machine learning. The most fundamental requirement of geological big data mining is to establish a set of deep neural networks through machine learning to predict changes in geobodies and underground mineral distribution based on the structure and attribute distribution data of geobodies. The attribute model enables the unified storage and management of the structure and attribute distribution data of geobodies, providing effective support for various analysis methods of geological big data.
Gridding serves as the technical foundation of attribute modeling, quantitative analysis, and numerical simulation. It is also the research focus of the new generation of 3D geological modeling technology. The grid type and gridding quality produce significant impacts on 3D geological modeling. 3D gridding faces two technical challenges. First, geologic conditions cannot be simplified during the 3D gridding of geobodies. Second, it is difficult to simplify the geological information when the 3D geological mesh is cut by man-made structures. Gridding can be achieved using most 3D geological modeling software, which, however, yields 3D mesh models using varying algorithms. This significantly affects the accuracy of subsequent analyses and calculations, determining whether the numerical simulation of solid mechanics can be performed. It is challenging to employ traditional gridding technologies to construct a 3D mesh that allows for accurate characterization of geological structures, as well as complex analyses and calculations. In contrast, the new 3D modeling method proposed in this study is more effective in gridding due to its integration of truncated rectangular, perpendicular bisection (PEBI) grids, and corner-point grids.
Truncated rectangular grids are irregular, allowing faults to be expressed using polygonal grids without losing their structural information. Besides, these grids are simple and efficient, supporting gridding under the constraint of complex structures.
PEBI grids are unstructured grids that remain smooth without serrations at the stratigraphic and fault levels. These grids can be densified around boreholes, leading to more accurate simulation results. PEBI grids also support gridding under the constraint of complex structures.
Corner-point grids are commonly used in the geological modeling and numerical simulation of oil reservoirs. However, these grids will lose structural information where strata pinch out or faults intersect. Besides, it is inapplicable to the local updating of a 3D geological model.
Rasters are simple regular grids characterized by simple spatial position expressions. They have a limited ability to express complex geobodies. Furthermore, they typically represent structural information using attribute interfaces, failing to express structures effectively.
Attribute modeling is to establish attribute models based on structural models by conducting calculations using a geostatistical method. The attributes involved consist of discrete attributes (e.g., sedimentary facies and lithofacies) and continuous attributes (e.g., water content, weight, permeability, and saturation). The commonly used interpolation algorithms include Kriging, sequential simulation, and inverse distance weighting (IDW).
Data analysis refers to data research and preprocessing using many methods before modeling. Attribute data for modeling can be classified into discrete data (for facies modeling) and continuous data (for physical property modeling), which correspond to varying analysis methods. The variogram function is a method to describe the variations in reservoir properties. This method follows the principle that samples in close proximity exhibit a stronger correlation than those farther apart. Furthermore, the further change in the distance, after exceeding a certain value, exerts minimal influence on the correlation. Discrete attributes can be directly analyzed using the variogram function. In contrast, continuous attributes should be transformed first. In this study, many transformation algorithms have been developed, including input and output truncation, logarithmic and exponential transformation, standardization, and 1D, 2D, and 3D trend removal algorithms. The variogram function analysis yields experimental variogram functions, which need to be fitted. Previous studies indicate four major theoretical variogram models for fitting (Fig. 4). Of these, the spherical model, the most widely used in geostatistics, can effectively determine many variables through fitting. A variogram function contains three pivotal parameters: sill, range, and nugget. Nugget C0 is the Y-intercept of the function, indicating the discontinuous change of a variable below the sampling scale. Sill C0+C is the maximum of the variogram function, representing the maximum variation in attributes. Range a is the minimum lag distance where the value of the function remains unchanged, indicating that the spatial correlation of the variable vanishes in the case of a lag distance greater than a. Therefore, range a reflects the influence range of the variable. With the continuous advancement in science and technology, the computing power of computers has been enhanced constantly, promoting the rapid development of numerical methods. These numerical methods can combine material and structural heterogeneity, supporting investigations into subsurface characteristics.
The X area, located in Southwest China, spans the Sichuan Basin and the Tibetan Plateau, exhibiting an extensive area and complex geologic conditions. In this area, 56 types of mineral resources (with proven reserves for 30 types) and 620 mining areas have been identified. This study aims to construct a high-precision, large-scale 3D model for the X area using available geological maps, borehole data, and airborne remote sensing data. This model is expected to support future breakthroughs in mineral exploration.
The 3D geological modeling of the X area was conducted using four angular coordinates and the new 3D geological modeling method. Based on theories of geology and other related disciplines, this study investigated and assessed the primary structural features within the area, including topographic features, soil-rock interfaces, lithologic boundaries, lithologic assemblages, faults, folds, and structural planes, as well as subsurface stratigraphic distribution and structural morphologies at certain depths. This contributes to a detailed understanding of the spatial distribution and changes of surface and subsurface geobodies in the X area. The new 3D modeling method was applied to establish a high-resolution holistic 3D geological model of the area through gridding and multiple attribute modeling. The technology roadmap is shown in Fig. 5.
Modeling was conducted using layer data (e.g., basic geological ages and formation data) from nearly 1960 boreholes with depths ranging from 300 m to 1600 m (Table 1).
| Borehole No. | Layer No. | Layer depth | Thickness | Age on the stratigraphic column | Formation | |
| From | To | |||||
| 13-2 | 1 | 0 | 1.6 | 1.6 | Ptd5-1-2-(三) | |
| 13-2 | 2 | 1.6 | 132.76 | 131.16 | ∈ | Jiujitan |
| 13-2 | 3 | 132.76 | 288.8 | 156.04 | ∈ | Yulongshan |
| 13-2 | 4 | 288.8 | 299.97 | 11.17 | ∈ | Yulongshan |
| 13-2 | 5 | 299.97 | 364.61 | 64.64 | ∈ | Changxing |
| ··· | ··· | ··· | ··· | ··· | ··· | ··· |
DownLoad:
CSV
Two 1:1000000 maps of the X area are available, and their splicing part was selected for modeling (red box in Fig. 6). Specifically, the modeling area covers an area of 31104 km2 along the margin of the Sichuan Basin, exhibiting significant terrain differences and complex structures. This area is dominated by deep-lying hills and mountains, with elevations ranging from 1000 m to 3000 m and relative elevation differences of around 1000 m.
Fig. 6 illustrates the Chamdo H-47 and Chengdu H-48 maps. The area covered by the Chamdo H-47 map is located at 96°6'E‒102°5'E and28°6'N‒32°6'N primarily in Chamdo, Tibet, with elevations ranging from 2296 m to 5460 m (average: about 3500 m). This area exhibits complex and diverse geological characteristics and contains relatively complete strata except for the pre-Sinian and Cambrian strata. The strata in this area show a significant regional distribution and distinct contact relationships. Minerals in this area are dominated by copper and iron.
The area covered by the Chengdu H-48 map is located at 102°54'E‒104 ° 53'E and 30°05'N‒31°26'N. This area is dominated by deep-lying hills and mountains, with elevations mostly ranging between 1000 m and 3000 m. The tectonic belt in this area primarily contains faults and folds, which significantly control the distribution and deformations of the strata within. Over 60 types of mineral resources have been discovered in Chengdu H-48, with primary minerals including iron, titanium and vanadium, copper, lead, zinc, aluminum, gold, silver, strontium, and rare earth.
High-precision remote sensing images of the X area were used for modeling, covering 93°6'E‒104°53'E and 28°30'N‒32°28'N (Fig. 7). They serve as a reference for modeling and the visualization of the surface integration of 3D models, supporting professional studies. Remote sensing images can objectively reflect the morphologies, structures, spatial relationships, and other features of surface scenes, providing macroscopic, comprehensive insights into surface features. A high spatial resolution is generally preferred (Sousa AMO., 2017; Sousa AMO et al., 2015). High-quality land cover data like high-resolution optical remote sensing images serve as critical data sources of surface information (Liu S et al., 2017; Amarsaikhan D et al., 2009; Xu H, 2008; Xiang D et al., 2016; Waqar MM et al., 2012). These data sources can provide true and objective records of surface scenes, significant for 3D terrain simulation.
XRF spectroscopy data were obtained through the analysis of core samples. These data provide detailed information about the elemental composition of the samples, including the types and concentrations of elements, playing an essential role in constructing accurate geological models. Hence, they are of critical application value in 3D geological modeling. As a complex process involving the precise description and visualization of subsurface geological structures, stratigraphic distribution, and mineral resources, 3D geological modeling necessitates the combination of multiple data sources and technical means. As an important data source, the XRF spectroscopy data can complement and verify other data, contributing to the construction of more accurate and comprehensive geological models. The visualization of 3D geological models allows for a more intuitive presentation of subsurface geological structures, thus offering valuable decision support for geological exploration and mineral exploitation. In this study, the XRF spectroscopy data of core samples from 200 boreholes were used (Table 2).
| Instrument S. N. | Sample ID | Depth | Unit | Al content | Si content | S content | K content | Fe content | Rb content |
| 801712 | Zk01—3—035 | 200.08 | PPM | 127033 | 616101 | 6012 | 148045 | 65667 | 215 |
| 801712 | zk01—3—036 | 201.13 | PPM | 14596 | 537842 | 18953 | 24991 | 8467 | 21 |
| 801712 | zk01—3—037 | 202.11 | PPM | 42030 | 663646 | 2333 | 53194 | 11432 | 56 |
| 801712 | zk01—3—038 | 203.07 | PPM | 11775 | 438962 | 73185 | 28308 | 10316 | 29 |
| 801712 | zk01—3—039 | 204.04 | PPM | 29899 | 574080 | 11807 | 35313 | 6600 | 37 |
| 801712 | zk01—3—040 | 205.08 | PPM | 16342 | 438455 | 36275 | 30014 | 9023 | 39 |
| 801712 | zk01—3—041 | 205.91 | PPM | 36549 | 596425 | 10264 | 46063 | 6315 | 51 |
| 801712 | zk01—3—042 | 206.82 | PPM | 72425 | 673585 | 2514 | 72728 | 20339 | 83 |
| 801712 | zk01—3—043 | 207.53 | PPM | 31436 | 466417 | 1259 | 36567 | 13436 | 52 |
| 801712 | zk01—3—044 | 208.57 | PPM | 54600 | 660911 | 2325 | 65139 | 11030 | 54 |
| 801712 | zk01—3—045 | 209.18 | PPM | 141864 | 562079 | 3358 | 174241 | 69117 | 254 |
| 801712 | zk01—3—046 | 211.06 | PPM | 134457 | 580616 | 3463 | 172020 | 51150 | 235 |
| 801712 | zk01—3—047 | 212.41 | PPM | 26300 | 595718 | 23573 | 54750 | 5977 | 54 |
| 801712 | zk01—3—048 | 213.42 | PPM | 110714 | 452093 | 3914 | 102366 | 43068 | 207 |
| 801712 | zk01—3—049 | 214.45 | PPM | 134738 | 582854 | 5527 | 165485 | 60418 | 241 |
| 801712 | zk01—3—050 | 215.47 | PPM | 101401 | 381163 | 5064 | 92111 | 56377 | 117 |
| ··· | ··· | ··· | ··· | ··· | ··· | ··· | ··· | ··· | ··· |
DownLoad:
CSV
The physical parameters of cores play a significant role in 3D geological modeling. These data provide detailed information about the physical properties of rocks, including their porosity, permeability, and density, which are essential for the construction of accurate and detailed 3D geological models. First, these data help determine the physical properties and distributions of rocks, contributing to more accurate simulations of the structures, morphologies, and properties of subsurface rock layers. By inputing these data into 3D modeling software, 3D geological models with actual physical properties can be constructed. Second, these data feature high accuracy and reliability since they are typically derived from core sample analysis. Accordingly, introducing these data can improve the accuracy and reliability of 3D geological models by reducing errors and uncertainties, leading to the creation of more practical models. Third, they can be used to optimize the parameter settings of 3D geological models. In the process of modeling, it is necessary to adjust the model parameters according to the actual situation to achieve more practical models that reflect geological laws. The physical parameters of cores provide a significant reference for parameter adjustment, contributing to the construction of more accurate and practical models. Finally, the visualization and quantitative analysis of subsurface complex geological structures and reservoirs can be achieved by combining these data with 3D modeling technology. In this study, the physical parameters of core samples from 200 boreholes were used (Table 3).
| Sample | Depth | Sericite | Sericite wavelength | IC | Cu | Mo | Au | Ag |
| ZK02_0043_7065 | 203.25 | 0.11 | 2208.59 | 0.675 | 0.42 | 59.161 | 1.04 | 0.06 |
| ZK02_0043_7066 | 203.27 | 0.108 | 2208.81 | 0.673 | 0.42 | 58.801 | 1.04 | 0.06 |
| ZK02_0043_7067 | 203.29 | 0.0976 | 2208.98 | 0.659 | 0.42 | 58.442 | 1.04 | 0.06 |
| ZK02_0043_7068 | 203.32 | 0.0974 | 2208.77 | 0.673 | 0.42 | 57.902 | 1.04 | 0.06 |
| ZK02_0043_7069 | 203.34 | 0.0969 | 2208.83 | 0.686 | 0.42 | 57.543 | 1.04 | 0.06 |
| ZK02_0043_7070 | 203.37 | 0.0838 | 2208.92 | 0.648 | 0.42 | 57.004 | 1.04 | 0.06 |
| ZK02_0043_7071 | 203.39 | 0.0912 | 2208.78 | 0.694 | 0.42 | 56.644 | 1.04 | 0.06 |
| ZK02_0043_7072 | 203.41 | 0.0626 | 2208.96 | 0.511 | 0.42 | 56.285 | 1.04 | 0.06 |
| ZK02_0043_7073 | 203.44 | 0.0641 | 2208.13 | NULL | 0.42 | 55.745 | 1.04 | 0.06 |
| ZK02_0043_7074 | 203.46 | 0.0518 | 2210.35 | 0.569 | 0.42 | 55.386 | 1.04 | 0.06 |
| ··· | ··· | ··· | ··· | ··· | ··· | ··· | ··· | ··· |
DownLoad:
CSV
The borehole data were organized in a uniform data format, with points of various layers arranged into three groups. These data were used to construct models of varying scales and precision to overcome overly thin layers within a too large working area. The coordinate system of the borehole location data was transformed from the World Geodetic System 1984 (WGS 84) to the China Geodetic Coordinate System 2000 (CGCS2000) using the Gauss- Krüger projection. Regarding the organization of geological maps, the 1∶100000 Chengdu and Chamdo maps were projected onto the CGCS2000 to ensure consistency between projection parameters and borehole data. For the organization of surface elevation data, 90-m-resolution data were downloaded from the Geospatial Data Cloud initially and were then clipped and sparsified. The raster data were transformed into point data, followed by projection onto the CGCS2000. The high alignment among datasets (Fig. 8) indicates correct projection.
To design a convolutional neural network (CNN) for geological data processing, it is necessary to carefully consider the characteristics of geologic data, modeling requirements, and computational resources. In this study, a CNN architecture was designed to process prime geologic data and identify and classify geobodies. This architecture consisted of multiple convolutional layers, pooling layers, fully connected layers, and activation functions, aimed at extracting features and classifying geologic data.
The input data of the CNN architecture included borehole data, physical parameters of cores, XRF spectrometry data, geological maps, and remote sensing images. These data were used to train the CNN model after preprocessing, typically represented as 2D or 3D matrices. Data preprocessing included data standardization and augmentation. Data standardization was performed to ensure that all feature values were within the same range, thereby accelerating training and enhancing model performance. Data augmentation was conducted using techniques like flipping, rotating, and cropping, aiming to increase data diversity and improve the model’s generalization ability. The input data were divided into the training, validation, and test sets, with the training set comprising 70 % to 80 % of the input data. The validation and test sets, each representing 10 % to 15 % of the input data, were used to tune model hyperparameters and assess final model performance, respectively.
The designed CNN framework consisted of an input layer, convolutional layers, pooling layers, fully connected layers, and an output layer. The input layer accepted geologic data as input, including borehole data, the physical parameters of cores, XRF spectrometry data, geological maps, and remote sensing images. The channel dimension and number in the input layer were adjusted according to the dimensions and resolution of the input data. Multiple convolutional layers were used to perform convolution operations on the input data to extract local features. Specifically, five convolutional layers were employed to extract features at different levels. A ReLU activation was applied after each convolutional layer to introduce nonlinearity. Pooling layers were inserted between convolutional layers to reduce both the dimensionality of feature maps and the computational load using average pooling. After convolution and pooling operations, three fully connected layers flattened the feature maps into a one-dimensional vector, followed by the gradual extraction of global features and classification. In the output layer, the number of neurons was determined based on the classification task. The Softmax activation function was used to output the probability distribution of each class.
In the design of the CNN architecture, the loss function was the focus of the training and optimization strategies. Based on the task requirements, the cross-entropy loss function was selected for classification. An appropriate Adam optimizer was used to update the network parameters, thereby accelerating the model’s training process. Additionally, L1 and L2 regularization was employed to prevent overfitting and enhance the model’s generalization performance. The following aspects were considered in the design. First, for the network structure design, it is crucial to balance the actual computational resources and time constraints to avoid the training difficulty and high computational costs caused by an overly complex network. Second, in the training process, close attention should be paid to the model’s convergence and overfitting issues, and a timely adjustment of the training strategy and network structure is required. The CNN framework designed in this study is illustrated in Fig. 9.
A 3D geological structural model of the X area was established using advanced 3D geological modeling methods. The data sources for modeling include borehole data, geological maps, and remote sensing images. Through multisource heterogeneous data processing, large-scale and high-precision modeling, gridding, and attribute modeling, a high modeling accuracy was achieved, with each grid representing an area of 50 m × 50 m on average. Considering the large scale of the modeling area, the large-scale, high-precision holistic modeling technology was employed. The modeling area was divided into four parts for independent modeling. Then, the resulting models were seamlessly integrated to form a comprehensive 3D geological model. Finally, the comprehensive model was demonstrated through OpenGL 3D rendering and the superimposition of oblique photogrammetry models. The constructed 3D geological model of the X area is shown in Fig. 10.
Under the constraint of the structural model, geobodies were divided into truncated rectangular grids. Such gridding can ensure the highly consistent geometric morphology between the geological attribute models and the structural model, as well as smooth grids at the stratigraphic and fault levels (Fig. 11). The resulting grid model is a multi-precision model that supports multi-resolution grids, rendering it well-suited for visualization in a level of detail (LOD) mode.
After spatial coordinate transformation, the large-scale high-precision geological model and the remote sensing image were fused seamlessly. Their perfect agreement suggests high modeling accuracy (Fig. 12).
The complex and irregular surface landscapes of the X area render it challenging to achieve satisfactory 3D surface morphologies using simple texture mapping. The realism of 3D terrain simulation can be significantly improved by pasting processed remote sensing images on the 3D terrain surface as textures (Bhaskaran S et al., 2010). Surface texture mapping that combines remote sensing images with DEMs emerges as an effective approach to enhance the realism of 3D terrain simulation. Most especially, this approach holds promising application prospects for the 3D landscape visualization of areas with inconvenient traffic and a sparse population (Bhaskaran S et al., 2010). Fig. 13 illustrates the 3D geological model combined with remote sensing images. This model enables interactions such as cutting and excavation. This facilitates geological studies and also provides a reference and basis for determining the rationality of the geological structures and the accuracy of sedimentary area division and geotechnical property description in the model.
The new 3D geological modeling method plays a crucial role in exploration breakthroughs. This method integrates complementary information from multiple data sources while also presenting geological structures and ore body distributions three-dimensionally, as exemplified by the ore body marked by a red box in Fig. 14. This will significantly enhance the accuracy and efficiency of exploration prediction.
The new 3D geological modeling method has played a critical role in delineating a metallogenic target area in a region in Southwest China, as denoted by the red box in Fig. 15. This method enables more accurate identification of key information, such as mineral deposit distribution and the morphologies and distributions of geological structures. Accordingly, potentially favorable structures subjected to mineralization can be determined more accurately. This contributes to reduced target area ranges, increased success rates of mineral exploration drilling, and enhanced discovery rates of new mineral deposits.
3D geological modeling is a complex process, and its achievements and cost depend on multiple factors such as the extent of the modeling area, the complexity of geological structures, and data types. For instance, the size of the 3D geological model is closely related to the extent of the modeling zone and the complexity of geological structures. A larger modeling area requires more data and more complex computations, resulting in larger model files. For the 3D geological modeling project in this study, the modeling area covers an area of 31104 km2, and a data volume of up to 972 GB is required. Furthermore, the model area is situated in the Sichuan Basin, exhibiting complex geological structures such as multiple faults, inclined strata, and magmatic intrusions.
Primary costs of 3D geological modeling include data acquisition, labor, and time costs. The data acquisition cost is related to the high-expense methods, such as drilling and satellite remote sensing. The labor cost is a significant cost component, as the involvement of professional geological engineers is essential for modeling. Moreover, complex modeling projects typically require collaboration among multidisciplinary teams. The time cost warrants much attention since the construction and optimization of a complex model generally take several months. Generally, the total cost of 3D geological modeling varies from hundreds of thousands to millions of yuan, depending on the project’s scale and complexity. Based on previous work experience, it was initially estimated that the 3D geological modeling in this study required three people, four months of work, and a budget of 200000 yuan. However, by applying the new 3D geological modeling method, the modeling was completed by only two people in two months, with a cost of merely 80000 yuan. The modeling process proves efficient due to data and algorithm optimization. The data optimization is attributed to advanced technology for multisource heterogeneous data processing. This technology converts heterogeneous data into homogeneous vector data (points and lines) to provide foundational surface data of ground, sections, and rock mass surfaces. The algorithm optimization includes parallel modeling, large-scale and high-precision modeling, seamless splicing of local models, and cluster analysis. All of these algorithm optimization measures can reduce manual intervention and repetitive tasks, thereby reducing the modeling time and labor cost. The data and algorithm optimization ensure that the new 3D geological modeling approach produces high-quality geological models while significantly reducing the model size and overall cost. Therefore, the 3D geological modeling method holds great value in geological engineering and resource exploration.
The structural modeling practice of the X area highlights the importance of the novel 3D geological modeling method, which can significantly enhance the effectiveness of geologic data application and the accuracy of 3D geological models. The future development of 3D geological modeling technology should focus primarily on two aspects. The first is to reduce modeling costs and improve modeling efficiency, and the second is to enhance model quality to meet user needs more comprehensively. By integrating remote sensing images and borehole data, the new modeling method can produce a complete and detailed 3D representation of the subsurface conditions nationwide. This will substantially facilitate the work of global researchers. The authors will continue to make technical and methodological improvements while closely monitoring relevant progress.
With the continuous development of human society and the increasing exploitation of surface mineral resources, major mining nations in the world have turned their attention to the more challenging deep mineral resource exploitation. Notably, 3D geological modeling technology can provide robust support for the exploration of deep mineral resources. With the progress in a new round of strategic mineral exploration initiatives, big data, and machine learning have demonstrated significant advantages in addressing challenges such as mineral resource localization and prediction. Therefore, deep geologic data mining by integrating big data, machine learning, and 3D visualization modeling might be an effective method to make key breakthroughs in mineral exploration. This is also the subsequent study focus of the authors.
Zhao-yang Ma, Sen Zhang, and Ya-kai Qiao conceived of the presented idea. Zhao-yang Ma and Hua-feng Sun developed the theory and performed the computations. Zhao-yang Ma, Ming Jing, and Jie Li verified the analytical methods. Zhao-yang Ma encouraged Fateh Bouchaala to supervise the findings of this work. All authors discussed the results and contributed to the final manuscript.
The authors declare no conflicts of interest.
This work was supported by a project of the China Geological Survey (DD20240126).
|
Amarsaikhan D, Ganzorig M, Blotevogel HH, Nergui B, Gantuya R. 2009. Integrated approach to extract information from high and very high resolution RS images for urban planning. Journal of Geography and Regional Planning. 2, 258– 267. doi: 10.5897/JGRP.
|
|
Bhaskaran S, Paramananda S, Ramnarayan M. 2010. Per-pixel and object-oriented classification methods for mapping urban features using Ikonos satellite data. Applied Geography, 30(4), 650–665. doi: 10.1016/j.apgeog.2010.01.009.
|
|
Biggs J, Ebmeier SK, Aspinall WP, Lu Z, Pritchard ME, Sparks RSJ, Mather TA. 2014. Global link between deformation and volcanic eruption quantified by satellite imagery. Nature Communications, 5, 3471. doi: 10.1038/ncomms4471.
|
|
Biggs J, Pritchard ME. 2017. Global volcano monitoring: What does it mean when volcanoes deform? Elements, 13(1), 17–22. doi: 10.2113/gselements.13.1.17.
|
|
Chen F, Wu YN. 2009. Three-dimensional visualization based on digital elevation model and remote sensing image. Science of Surveying and Mapping, 34(S2), 127–129 (in Chinese).
|
|
Du XF, Feng W, Yang QX. 2018. The supervised classification of lithology based on ZY-3 image. Resources Environment & Engineering, 032, 291–295. doi: 10.16536/j.cnki.issn.1671-1211.2018.02.027.
|
|
Estoque RC, Murayama Y. 2015. Classification and change detection of built-up lands from Landsat-7 ETM+ and Landsat-8 OLI/TIRS imageries: A comparative assessment of various spectral indices. Ecological Indicators, 56, 205–217. doi: 10.1016/j.ecolind.2015.03.037.
|
|
Fernandez J, Prieto JF, Escayo J, Camacho AG, Luzón F, Tiampo KF, Palano M, Abajo T, Pérez E, Velasco J, Herrero T, Bru G, Molina I, López J, Rodríguez-Velasco G, Gómez I, Mallorquí JJ. 2018. Modeling the two- and three-dimensional displacement field in Lorca, Spain, subsidence and the global implications. Scientific Reports, 8, 14782. doi: 10.1038/s41598-018-33128-0.
|
|
Ge W, Yang H, Zhu XB, Ma MG, Yang YL. 2018. Ghost city extraction and rate estimation in China based on NPP-VIIRS night-time light data. ISPRS International Journal of Geo-Information, 7(6), 219. doi: 10.3390/ijgi7060219.
|
|
Hao M, Wu H, Jia ZQ, Huang DN, Liu Y, Guan Z, 2014. Discussion on genetic Correlation between Huashan rock mass and Gupaishan rock mass based on comparative analysis of remote sensing image characteristics of fault structure. Remote Sensing for Natural Resources, 26, 162–169. doi: https://doi.org/10.6046/gtzyyg.2014.02.26.
|
|
He HJ, Bai SW, Zhao XH, Cheng J. 2002. Discussion on strata partition in three dimension strata model. Rock and Soil Mechanics, 23(5), 637–639 (in Chinese with English abstract). doi: 10.16285/j.rsm.2002.05.026.
|
|
Horning N. 2008. Remote sensing. Encyclopedia of Ecology. Amsterdam: Elsevier, 2986–2994. doi: 10.1016/b978-008045405-4.00237-8.
|
|
Huang Y, Yu M, Xu Q, Sawada K, Moriguchi S, Yashima A, Liu CW, Xue L. 2015. InSAR-derived digital elevation models for terrain change analysis of earthquake-triggered flow-like landslides based on ALOS/PALSAR imagery. Environmental Earth Sciences, 73(11), 7661–7668. doi: 10.1007/s12665-014-3939-5.
|
|
Jiang W, He GJ, Long TF, Wang C, Ni Y, Ma RQ. 2017. Assessing light pollution in China based on nighttime light imagery. Remote Sensing, 9(2), 135. doi: 10.3390/rs9020135.
|
|
Kenduiywo BK, Tolpekin VA, Stein A. 2014. Detection of built-up area in optical and synthetic aperture radar images using conditional random fields. Journal of Applied Remote Sensing, 8(1), 083672. doi: 10.1117/1.jrs.8.083672.
|
|
Levin N, Duke YS. 2012. High spatial resolution night-time light images for demographic and socio-economic studies. Remote Sensing of Environment, 119, 1–10. doi: 10.1016/j.rse.2011.12.005.
|
|
Li CL, Li FD, Li JQ, Liu YY, Liu C, Lv X. 2015. Smart geological survey architecture. Geology in China, 42(4), 828–838 (in Chinese with English abstract). doi: 10.3969/j.issn.1000-3657.2015.04.003.
|
|
Li CL, Zhang H, Zhu LF. 2013. Algorithm for true-3D modeling of geological body with single-fault. Computer Engineering and Design, 34(10), 3590–3594 (in Chinese with English abstract). doi: 10.16208/j.issn1000-7024.2013.10.046.
|
|
Li JC, Wu ZH, Zhang D, Liu XD. 2014. Remote sensing image interpretation and tectonic activity study of the main active faults in Yushu area, Qinghai Province. Geological Bulletin of China, 33(4), 535–550 (in Chinese with English abstract). doi: 10.3969/j.issn.1671-2552.2014.04.010.
|
|
Li KN, Chen YH. 2018. A genetic algorithm-based urban cluster automatic threshold method by combining VIIRS DNB, NDVI, and NDBI to monitor urbanization. Remote Sensing, 10(2), 277. doi: 10.3390/rs10020277.
|
|
Li XC. 2005. Multi-Source remote sensing image fusion technology and application research. PLA University of Information Engineering
|
|
Li XM, Zhou WQ. 2018. Dasymetric mapping of urban population in China based on radiance corrected DMSP-OLS nighttime light and land cover data. Science of the Total Environment, 643, 1248–1256. doi: 10.1016/j.scitotenv.2018.06.244.
|
|
Liu SC, Tong XH, Bruzzone L, Du PJ. 2017. A novel semisupervised framework for multiple change detection in hyperspectral images. 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). July 23-28, 2017, Fort Worth, TX, USA. IEEE, 173–176. doi: 10.1109/IGARSS.2017.8126922.
|
|
Liu YJ, Li M, Song LJ, Wang Z. 2012. Analysis on three-dimensional orebody data model based on the object-oriented techniques. Journal of Liaoning Technical University (Natural Science), 31(4), 437–440 (in Chinese). doi: 10.3969/j.issn.1008-0562.2012.04.002.
|
|
Ma XL, Tong XH, Liu SC, Ma ZT. 2017. Extraction of built-up areas in Chinese silk road economic belt based on DMSP-OLS data. 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). July 23-28, 2017, Fort Worth, TX, USA. IEEE, 5877–5880. doi: 10.1109/IGARSS.2017.8128346.
|
|
Ma ZY, Wang ZC, Zhang LH, Yao YT, Qiao YK. 2022. A new 3D geological modeling method and its application in Chengdu-Changdu region modeling. Northwestern Geology, 55(2), 82–92 (in Chinese with English abstract). doi: https://doi.org/10.19751/j.cnki.61-1149/p.2022.02.006.
|
|
Malenovský Z, Rott H, Cihlar J, Schaepman ME, García-Santos G, Fernandes R, Berger M. 2012. Sentinels for science: Potential of Sentinel-1, -2, and-3 missions for scientific observations of ocean, cryosphere, and land. Remote Sensing of Environment, 120, 91–101. doi: 10.1016/j.rse.2011.09.026.
|
|
Ouyang Y, Liu H, Li GM, Ma DF, Zhang LK, Huang HX, Zhang JH, Zhang TJ, Liu X, Zhao YB, Li F. 2023. Mineral search prediction based on random forest algorithm——A case study on porphyry-epithermal copper polymetallic deposits in the western Gangdise metallogenic belt. Geology in China, 50(2), 303–330 (in Chinese with English abstract). doi: 10.12029/gc20201026001.
|
|
Pesaresi M, Gerhardinger A, Kayitakire F. 2008. A robust built-up area presence index by anisotropic rotation-invariant textural measure. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 1(3), 180–192. doi: 10.1109/JSTARS.2008.2002869.
|
|
Slonecker ET, Jennings DB, Garofalo D. 2001. Remote sensing of impervious surfaces: A review. Remote Sensing Reviews, 20(3), 227–255. doi: 10.1080/02757250109532436.
|
|
Song JC, Tong XY, Wang LZ, Zhao CL, Prishchepov AV. 2019. Monitoring finer-scale population density in urban functional zones: A remote sensing data fusion approach. Landscape and Urban Planning, 190, 103580. doi: 10.1016/j.landurbplan.2019.05.011.
|
|
Sousa AMO, Gonçalves AC, Silva JRM. 2017. Above-ground biomass estimation with high spatial resolution satellite images. In: Tumuluru JS (ed. ) Biomass Volume Estimation and Valorization for Energy. InTech: Rijeka, Croatia, Volume 2017, 47–70. doi: 10.5772/65665.
|
|
Sousa AMO, Gonçalves AC, Mesquita P, Marques da Silva JR. 2015. Biomass estimation with high resolution satellite images: A case study of Quercus rotundifolia. ISPRS Journal of Photogrammetry and Remote Sensing, 101, 69–79. doi: 10.1016/j.isprsjprs.2014.12.004.
|
|
Stevens FR, Gaughan AE, Linard C, Tatem AJ. 2015. Disaggregating census data for population mapping using random forests with remotely-sensed and ancillary data. PLoS One, 10(2), e0107042. doi: 10.1371/journal.pone.0107042.
|
|
Wang DH. 2016. A discussion on some problems concerning deep exploration of mineral resources in South China. Geology in China, 43(5), 1585–1598(in Chinese with English abstract). doi: https://dx. doi.org/10.12029/gc20160509.
|
|
Wang KL. Construction of DEM at the bottom of Baiyangdian Wetland Based on Remote Sensing Thematic Interpretation. 2019. Modern Geology, 033(005), 1098–1105. doi: https://doi.org/10.19657/j.geoscience.1000-8527.2019.006.
|
|
Wang RS. 2011. Remote sensing geological exploration technology and application research. Acta Geographica Sinica, 85, 1699–1743
|
|
Wang Y, Wang DH, Wang YL, Huang F. 2021. Quantitative research on spatial distribution of antimony deposits in China based on geological big data. Geology in China, 48(1), 52–67 (in Chinese with English abstract). doi: 10.12029/gc20210104.
|
|
Waqar MM, Mirza JF, Mumtaz R, Hussain E. 2012. Development of new indices for extraction of built- up area and bare soil from Landsat Data. Open Access Scientific Reports. 1, 1–4.
|
|
Wasklewicz T, Staley DM, Reavis K, Oguchi T. 2013.3. 6 digital terrain modeling. Treatise on Geomorphology. Amsterdam: Elsevier, 130–161. doi: 10.1016/b978-0-12-374739-6.00048-8.
|
|
Woodcock CE, Allen R, Anderson M, Belward A, Bindschadler R, Cohen W, Gao F, Goward SN, Helder D, Helmer E, Nemani R, Oreopoulos L, Schott J, Thenkabail PS, Vermote EF, Vogelmann J, Wulder MA, Wynne R. 2008. Free access to landsat imagery. Science, 320(5879), 1011. doi: 10.1126/science.320.5879.1011a.
|
|
Xiang DL, Tang T, Hu CB, Fan QH, Su Y. 2016. Built-up area extraction from PolSAR imagery with model-based decomposition and polarimetric coherence. Remote Sensing, 8(8), 685. doi: 10.3390/rs8080685.
|
|
Xu H. 2008. A new index for delineating built-up land features in satellite imagery. International Journal of Remote Sensing, 29(14), 4269–4276. doi: 10.1080/01431160802039957.
|
|
Yu M, Huang Y, Zhou JM, Mao LY. 2017. Modeling of landslide topography based on micro-unmanned aerial vehicle photography and structure-from-motion. Environmental Earth Sciences, 76(15), 520. doi: 10.1007/s12665-017-6860-x.
|
|
Yu RQ, Gao JG, Zhao XL. 2013. Metallogenic prognosis of polymetallic lead-zinc deposits in Balao area of Lancang, Yunnan Province, based on comprehensive ore-forming information. Geology in China, 40(3), 967–973 (in Chinese with English abstract). doi: 10.3969/j.issn.1000-3657.2013.03.026.
|
|
Zhang WM, Qi JB, Wan P, Wang HT, Xie DH, Wang XY, Yan GJ. 2016. An easy-to-use airborne LiDAR data filtering method based on cloth simulation. Remote Sensing, 8(6), 501. doi: 10.3390/rs8060501.
|
|
Zhang Y, Zhang TF, Cheng XY, Sun LX, Cheng YH, Wang SY, Wang SB, Ma HL, Lu C. 2022. A brief analysis on the three-dimensional geological structure and uranium mineralization of Jurassic uranium-bearing rock series in the northeastern Ordos Basin. Geology in China, 49(1), 66–80 (in Chinese with English abstract). doi: https://dx. doi.org/10.12029/gc20220105.
|
|
Zhang ZH, Hou EK, Luo XX, Deng ND. 2014. Intersected model between tunnel and strata. Geomatics and Information Science of Wuhan University, 39(4), 496–499 (in Chinese with English abstract). doi: 10.13203/j.whugis20120507.
|
|
Zhou F, Wang BD, He J, Hao M, Wang P. 2022. 3D visualization modeling and application study of porphyry-skarn gold-copper deposits in Beiya Area, Yunnan Province. Geology in China, 49(1) 241–252 (in Chinese with English abstract). doi: https://doi.org/10.12029/gc20220115.
|
No. 45 Fuwai Street, Xicheng District, Beijing 100037, P. R. China
Supported by Beijing Renhe Information Technology Co. Ltd Technical support: info@rhhz.net
| Borehole No. | Layer No. | Layer depth | Thickness | Age on the stratigraphic column | Formation | |
| From | To | |||||
| 13-2 | 1 | 0 | 1.6 | 1.6 | Ptd5-1-2-(三) | |
| 13-2 | 2 | 1.6 | 132.76 | 131.16 | ∈ | Jiujitan |
| 13-2 | 3 | 132.76 | 288.8 | 156.04 | ∈ | Yulongshan |
| 13-2 | 4 | 288.8 | 299.97 | 11.17 | ∈ | Yulongshan |
| 13-2 | 5 | 299.97 | 364.61 | 64.64 | ∈ | Changxing |
| ··· | ··· | ··· | ··· | ··· | ··· | ··· |
DownLoad:
CSV
| Instrument S. N. | Sample ID | Depth | Unit | Al content | Si content | S content | K content | Fe content | Rb content |
| 801712 | Zk01—3—035 | 200.08 | PPM | 127033 | 616101 | 6012 | 148045 | 65667 | 215 |
| 801712 | zk01—3—036 | 201.13 | PPM | 14596 | 537842 | 18953 | 24991 | 8467 | 21 |
| 801712 | zk01—3—037 | 202.11 | PPM | 42030 | 663646 | 2333 | 53194 | 11432 | 56 |
| 801712 | zk01—3—038 | 203.07 | PPM | 11775 | 438962 | 73185 | 28308 | 10316 | 29 |
| 801712 | zk01—3—039 | 204.04 | PPM | 29899 | 574080 | 11807 | 35313 | 6600 | 37 |
| 801712 | zk01—3—040 | 205.08 | PPM | 16342 | 438455 | 36275 | 30014 | 9023 | 39 |
| 801712 | zk01—3—041 | 205.91 | PPM | 36549 | 596425 | 10264 | 46063 | 6315 | 51 |
| 801712 | zk01—3—042 | 206.82 | PPM | 72425 | 673585 | 2514 | 72728 | 20339 | 83 |
| 801712 | zk01—3—043 | 207.53 | PPM | 31436 | 466417 | 1259 | 36567 | 13436 | 52 |
| 801712 | zk01—3—044 | 208.57 | PPM | 54600 | 660911 | 2325 | 65139 | 11030 | 54 |
| 801712 | zk01—3—045 | 209.18 | PPM | 141864 | 562079 | 3358 | 174241 | 69117 | 254 |
| 801712 | zk01—3—046 | 211.06 | PPM | 134457 | 580616 | 3463 | 172020 | 51150 | 235 |
| 801712 | zk01—3—047 | 212.41 | PPM | 26300 | 595718 | 23573 | 54750 | 5977 | 54 |
| 801712 | zk01—3—048 | 213.42 | PPM | 110714 | 452093 | 3914 | 102366 | 43068 | 207 |
| 801712 | zk01—3—049 | 214.45 | PPM | 134738 | 582854 | 5527 | 165485 | 60418 | 241 |
| 801712 | zk01—3—050 | 215.47 | PPM | 101401 | 381163 | 5064 | 92111 | 56377 | 117 |
| ··· | ··· | ··· | ··· | ··· | ··· | ··· | ··· | ··· | ··· |
DownLoad:
CSV
| Sample | Depth | Sericite | Sericite wavelength | IC | Cu | Mo | Au | Ag |
| ZK02_0043_7065 | 203.25 | 0.11 | 2208.59 | 0.675 | 0.42 | 59.161 | 1.04 | 0.06 |
| ZK02_0043_7066 | 203.27 | 0.108 | 2208.81 | 0.673 | 0.42 | 58.801 | 1.04 | 0.06 |
| ZK02_0043_7067 | 203.29 | 0.0976 | 2208.98 | 0.659 | 0.42 | 58.442 | 1.04 | 0.06 |
| ZK02_0043_7068 | 203.32 | 0.0974 | 2208.77 | 0.673 | 0.42 | 57.902 | 1.04 | 0.06 |
| ZK02_0043_7069 | 203.34 | 0.0969 | 2208.83 | 0.686 | 0.42 | 57.543 | 1.04 | 0.06 |
| ZK02_0043_7070 | 203.37 | 0.0838 | 2208.92 | 0.648 | 0.42 | 57.004 | 1.04 | 0.06 |
| ZK02_0043_7071 | 203.39 | 0.0912 | 2208.78 | 0.694 | 0.42 | 56.644 | 1.04 | 0.06 |
| ZK02_0043_7072 | 203.41 | 0.0626 | 2208.96 | 0.511 | 0.42 | 56.285 | 1.04 | 0.06 |
| ZK02_0043_7073 | 203.44 | 0.0641 | 2208.13 | NULL | 0.42 | 55.745 | 1.04 | 0.06 |
| ZK02_0043_7074 | 203.46 | 0.0518 | 2210.35 | 0.569 | 0.42 | 55.386 | 1.04 | 0.06 |
| ··· | ··· | ··· | ··· | ··· | ··· | ··· | ··· | ··· |
DownLoad:
CSV
DownLoad: