Identification Of Geochemical Elements Distribution Patterns Based On Geostatistical Simulation,Fractal Topography And Singularity Analysis

Identification and extraction of metallogenic and mineralization information is a crucial link in mineral exploration.The metallogenic information represents the control mechanism of the host rock,ore-forming geophysics,and ore-controlling structure on the mineralization of the ore deposit.The mineralization information,which reflects the mineralization degree of interest regions,generally refers to the anomaly information extracted from geochemical,geophysical,remote sensing,and other exploration data.The geochemical exploration data clearly indicates where the ore-forming elements are concentrated in the study area.To improve the accuracy of metallogenic prediction,how to efficiently carry out the identification and anomaly extraction of geochemical elements distribution patterns from geochemical exploration data,and mine the deep-level mineralization information therein is crucial.The present position in the identification and anomaly extraction of geochemical elements distribution patterns mainly involves the construction of the spatial distribution of geochemical elements,the identification and evaluation of weak geochemical anomalies.The construction of a reliable geochemical element spatial distribution is the prerequisite for the identification of weak anomalies and the extraction of favorable mineralization information.In practical applications,deterministic interpolation techniques(such as IDW and Kriging)are useful in expressing the spatial correlation and variability of geochemical elements distribution.However,it will generate smoothed data to varying degrees,making it challenging to capture the complex structure of element spatial distribution and extract the mineralization information contained in the local singular value.In comparison,the spatial distribution characteristics and local variability of the raw data can be preserved to a significant extent by geostatistical stochastic simulation methods,which are helpful to characterize the local structural properties of element spatial distribution.It is acknowledged that the geochemical elements distribution is mostly the superposition of composite fields caused by different geological processes.The traditional geostatistical stochastic simulations have limitations in the quantification of its multimodal and complex patterns.Multiple-point geostatistics(MPS)method has been proven to have a better advantage in expressing complicated spatial structures and reconstructing target geometry compared with conventional methods.However,previous researches pay very little attention on the application of multiple-point geostatistical simulation in mineral potential mapping.Meanwhile,a single scale raster image was still utilized as the training image of MPS when constructing the geochemical element distribution in some pertinent researches.Additionally,there are still certain issues like the insufficient number of conditional points for simulation and the tiny number of data events gleaned from a single training image,resulting in the loss of local information during downscaling simulation.Therefore,how to fully utilize multi-scale and multi-type geochemical exploration data collected from the same area and realize the accurate characterization of the complicated spatial patterns of element distribution by MPS,has been a hot topic in the field of exploration geochemistry.According to the scale invariance and self-similarity of geochemical elements spatial pattern,the nonlinear model based on fractal and multifractal theory can quantitatively characterize the heterogeneity and singularity of elements distribution.The power-law relationship between density and measurement scale is the foundation for the development of anomaly identification methods based on fractal and multifractal theory.It reveals the law of complex phenomena changing with scale,which provides an effective tool for the extraction of weak geochemical anomaly,decomposition of composite or superposition information and quantitative description of spatial anisotropy of anomaly distribution.The traditional local singularity analysis method typically uses the average concentration of elements in the sliding window at different sizes to get the singularity index.However,the correlational researches showed that the size and geometric structure of sliding window have a great impact on the element spatial pattern,leading to enhanced local uncertainty.And the fixed window geometry is difficult to reflect the variation or anisotropic characteristics of element concentrations along different directions.Fractal topography gives the essential definition of scale invariance,which pointes out that the size of fractal dimension is controlled by fractal behavior.And the recently proposed multifractal topography offered a solid foundation for distinguishing between monofractals and multifractals,and clarified the physical meaning of the singularity index.The efficient combination of fractal topography and singularity analysis,especially giving full play to the advantages of fractal topography in describing the control mechanism of fractal properties,is meaningful for improving the recognition effect of mineralization anomaly information.This thesis focused on the identification of geochemical elements distribution patterns associated with the mineralization information of polymetallic deposits in light of these issues.On the basis of the scale invariance and self-similarity of geochemical elements spatial distribution,MPS,fractal topography,and singularity theory were used as the primary analytical tools to carry out the construction of geochemical elements distribution,the extraction of ore-forming elements combination,the identification and extraction of anisotropic geochemical anomalies,and the integration of multi-element geochemical anomaly information,respectively,hoping to provide a new solution for the identification of complex geochemical spatial pattern and anomaly extraction.To be specific,the main research contents and conclusions are as follows:(1)The applicability of commonly used spatial modeling methods in the construction of geochemical elements distribution was discussed in the dissertation.Aiming at the features of diverse scale types and heterogeneous distribution of geochemical data,a multiple-point geostatistical simulation method based on multi-scale geochemical exploration data was proposed to reproduce the complex pattern of geochemical elements distribution.Three steps make up the MPS method:the integration of geochemical exploration data from different scales and generation of simulation grid,the construction of multi-scale geochemical training image,and the downscaling simulation by direct sampling.Three key technologies are used to implement these steps:rasterization of sampling point data,fractal filtering to separate anomaly from background,and deterministic interpolation to fit the trend field.The fractal filtering method,such as S-A model,can successfully remove the background from the raster image to extract the complex pattern,while the deterministic interpolation technique can convert the geochemical data into the raster image and reproduce the changing trend of element distribution.The anomaly images of various types and scales after fractal filtering are used as the simulated grid and training images in direct sampling,as well as the background as the overall trend component,to realize the trend-based characterization of fine-scale geochemical elements distribution.Then,the geochemical elements distribution of ore-forming elements in the Zhongdian Island arc copper polymetallic deposit in Yunnan Province and the Wulonggou-Balong gold polymetallic deposit in Qinghai Province were simulated.The two case studies demonstrated that the MPS method proposed here is feasible and effective.It enhances the intensity of geochemical anomalies in local regions,improves the accuracy of the reproduction of spatial distribution patterns with similar trends,and has a good suitability for the spatial modeling of multi-scale geochemical exploration data.(2)The principle and limitation of the moment method in multifractal analysis were discussed,and a novel method of multifractal spectrum based on fractal topography was proposed to analyze the multifractal characteristics of element spatial distribution.This method mainly calculates the singularity index by using the relative ratio of the whole concentration to the local concentration as the basic statistic,to describe the relative enrichment and depletion of elements in local areas.De Wijs model,as an idealized model of multifractal,was used here to reveal the similarities and differences between the two methods.The result showed the basic consistency of the curve shapes of the two multifractal spectra,demonstrating the validity of the multifractal analysis method based on fractal topography.Compared with the moment method,it has a unified judgment criterion for the meaning of singularity index in different dimensional spaces,and the calculating process is simplified.Furthermore,the effects of staged factor analysis,multifractal analysis and ROC analysis in element combination extraction were discussed.The findings demonstrated that the integrated method considers the structural characteristics of component data,as well as the multifractal characteristics of the geochemical element distribution and its spatial correlation with the distribution of ore occurrences,which has been an important tool for the efficient extraction of mineralization indicator elements.(3)The problems of local singularity analysis in characterizing the anisotropic geochemical anomaly distribution were investigated,focusing on how to analyze anisotropic singularity based on the definition of self-affine fractal behavior in the generalized fractal topography.On this foundation,the parameter P_c was introduced to quantitatively describe the element depletion and enrichment state affected by the spatial range and geometry of the sliding window.The case study showed that the anomaly spatial distribution characteristics of elements in local areas affected by geological features of different ore-controlling factors,were effectively expressed through geochemical anomaly distribution patterns derived from various fractal topological relationships.Additionally,it was confirmed that the C-A model and P-A method are valid for analyzing the spatial correlation of ore occurrences and the spatial distribution pattern of geochemical anomalies.Then,the uncertainty analysis of geochemical anomaly distributions from different fractal topological relationships was conducted to quantitatively evaluate the risk of anisotropic geochemical distribution patterns on mineral exploration.(4)The fuzzy logic operators,expected value function,and multi-scale geographically weighted regression(MGWR)model were applied to identify multi-element geochemical anomaly signatures,which make up for the deficiency of single-element anomaly distribution pattern as the evidence layer of mineralization information.The predicting outcomes indicated that the expected value function and MGWR model were superior to the fuzzy logic operators in establishing the spatial relationship between the multi-element geochemical anomalies of the deposit-type sought and known ore occurrences.And the map of the expected value function shows a higher proportion of ore occurrences in high anomaly areas than that of MGWR model.In summary,the main contributions of this thesis can be concluded as follows:it was able to accurately characterize the fine-scale geochemical elements distribution using the multi-point geostatistical simulation,based on the complexity pattern extracted from multi-scale geochemical data;it proposed a set of methods for analyzing the multifractal and anisotropic characteristics of geochemical elements distribution patterns;it developed an integrated approach for the recognition of enhanced multi-element geochemical anomaly based on the singularity index.