User Login


ranking list


WeChat Public Number


Current Directory


  • Discussion on traditional research and big data research
    ZHANG Qi, ZHU Yueqin and JIAO Shoutao
    2019  .  38(12):    1939-1942    [摘要](88)    [PDF](68)
  • The deficiencies and possible solutions of TAS classification in the context of big data
    ZHANG Qi, GE Can, JIAO Shoutao, YUAN Feng, ZHANG Mingming and LIU Huiyun
    The TAS diagram, which is widely used in academia, is a classification scheme for volcanic rocks approved by IUGS in 1989. This classification has been active in advancing the research on petrology and geochemistry, regulating the naming of rocks (especially volcanic rocks) and facilitating the exchange of academia. However, it cannot be denied that earlier studies had some limitations due to analytical methods, analytical techniques and data volume limitations. Now that the world has accumulated a huge amount of data, researchers should have the conditions for making new discussion on the classification of volcanic rocks. In this study, the authors have developed a new TAS classification scheme for volcanic rocks using a probability density function. The original TAS diagram defines 15 root names, among which 9 are preserved in this paper. With the addition of one new root name, the authors define a total of 10 root names. The biggest change in the new TAS diagram is in the alkaline rock series. In the new TAS diagram, the trachybasalt series is closed up, thus making up for the earlier TAS classification in the acidic rock part of the deficiency. The authors have also found that the TAS diagram may have some problems. The classification scheme presented in this study is subject to further discussions and tests.
    2019  .  38(12):    1943-1954    [摘要](101)    [PDF](78)
  • Alkaline rock and the distinction between alkaline and sub-alkaline: A discussion on data of global volcanic rocks
    JIAO Shoutao, ZHANG Qi, GE Can, ZHU Yueqin, YUAN Lingling, SHAO Baorong, LING Xiao, YUAN Feng and ZHOU Yongzhang
    What is the true meaning of commonly used alkaline and sub-alkaline curves? It is reasonable for the International Union of Geological Sciences to make the requirement that alkaline rocks must have actual alkaline minerals and/or feldspar appearance as markers. However, the International Union of Geological Sciences also agrees that the basalt in area B of the TAS diagram could be subdivided into alkaline basalt and sub-alkaline basalt, depending on whether they have standard mineral nepheline. The basalt of the standard mineral nepheline is classified as alkaline basalt, while the Hawaiian rock is the alkaline basalt, due to the existence of standard minerals of nepheline. Why do they appear in the two districts on the TAS diagram (alkaline basalt in area B, Hawaiian Rock in the S1 area), and is this a conceptual confusion? It seems that the previous academic definition of alkaline and alkaline rock is somewhat confusing, the alkaline-sub-alkaline boundary determined by many experts is not the tholeiite and alkaline series, but is between normal series and dolerite series. In this paper, the global volcanic rock data are used to investigate TAS diagram and alkaline-sub-alkaline boundary, and it is found that alkaline and alkaline rocks are defective in the interpretation of terms and thus this paper proposes solutions. This matter involves the basic theory of petrology, to which the academia should pay more attention.
    2019  .  38(12):    1955-1962    [摘要](64)    [PDF](61)
  • The establishment of oceanic andesites tectonic environment discrimination diagrams with big data method.
    LIU Xinyu, ZHANG Qi and ZHANG Chengli
    Geochemical elements of magmatic rocks often indicate their tectonic environments. Previous geologists used tectonic environment discriminant diagrams to describe their correlation. However, it is too challenging to apply discriminant diagrams to identifying the tectonic environment of andesites because of their complexity of petrogenesis and the unicity of their tectonic environment. Based on the GEOROC and PetDB databases, the authors intergrated the global Cenozoic oceanic andesites with three categories:mid-oceanic ridge andesites (MORA), oceanic island andesites (OIA) and island arc andesites (IAA). With 924 element ratios consisting of any two of 43 elements, the authors built more than 420,000 rectangular coordinate systems. 4 optimal discriminant diagrams were sifted by calculating overlap ratios among the three types of oceanic andesites:lg(Ba/Nb) versus lg(Ga/Cs), lg(Eu/Pb) versus lg(TFeO/Ga), lg(Ga/Cs) versus lg(K2O/Nb) and lg(Cs/Nb) versus lg(MnO/Pb). The elements and element ratios were analyzed by comparing the kernel densities of the three types of andesites, with some conclusions reached:(1) The ratio of LILE and HFSE can effectively differentiate MORA and IAA; (2) the ratio of LILE and other elements is useful to identifying OIA from the other two types; (3) in a certain degree, LILE is more appropriate for determining tectonic environments of oceanic andesites than HFSE. This study presents that andesite is likely to be a widely used indicator of tectonic environments, which might be more appropiate than basalt discriminant diagram. It further indicates that even the andesite genesis is much more complicated than basalt, big data method is an effective approach to extract the correlation with tectonic discriminant significant.
    2019  .  38(12):    1963-1970    [摘要](71)    [PDF](79)
  • Difference between komatiites and picrites and a discussion on some Late Paleozoic “komatiites”
    LI Zhenhuan, LIU Xuelong, ZHU Yueqin, ZHANG Qi, LUO Ying, ZHANG Changzhen, CHEN Jianhang, WANG Shuaishuai and YANG Fucheng
    In the past, academia paid much attention to the similarity between komatiites and picrites, but ignored their differences. In this paper, the global data of Archaean komatiites and Post-Archaean low/high titanium picrites in the database were collected by full data model. Based on comparing the differences between them, the authors found that komatiites are richer in MgO, Cr, Ni, Cs, Pb, Co and Zn, followed by low-titanium picrites (except for Co and Zn). As for the other main and trace elements, high-titanium picrites has the highest content, followed by low-titanium picrites and then by komatiites. Based on the differences between elements such as Cr/Ga, MgO/Ga, MnO/Zr and Cr/Zr, the authors used density distribution to draw an isodensity discriminant map which can effectively distinguish the three types of rocks, and redefined the lithology of some Late Paleozoic "komatiites" with this diagram. The results of lithofacies and geochemical characteristics show that, in the Late Paleozoic "komatiites", the rocks in the eastern part of India are high-titanium picrites, those in Vietnam are low-titanium picrites with similar chemical composition to komatiites, and those in Ladak area of India are low-titanium picrites.
    2019  .  38(12):    1971-1980    [摘要](60)    [PDF](61)
  • A comparison of tree-based ensemble algorithms on the main element content of monoclinal pyroxene in mafic-ultramafic rocks
    SUN Jiankun, DU Xueliang, ZHANG Baoyue, WANG Long, JIN Weijun, ZHANG Qi, LUO Xiong and ZHU Yueqin
    Relying on the geochemical composition of the magma tectonic environment to understand the formation process of magma is an important application in rock geochemistry. While the current works to make full use of rock geochemical components for the tectonic setting discrimination are not enough. In this study, the authors utilized four tree-based machine learning methods to make magma tectonic environment discriminations and feature sorting on the 13 main ingredients of monoclinal pyroxene in maficultramafic rocks from global Cenozoic ocean island (OIB), island arc (IAB), and mid-ocean ridge (MORB). Through the comparison of the four tree-based machine learning methods, the authors proved the validity of the tree-based methods for the identification of geochemical components and derived the advantages and disadvantages of the four methods in dealing with the identification of rock tectonic environments:decision trees gain better comprehensibility but have lower recognition accuracy, boosting algorithms AdaBoost and GBDT have the best recognition accuracy but lower comprehensibility, and random forest is a better choice during trading off and comprehensibility performance. Besides, Cr2O3, TFeO, TiO2, FeO and Al2O3 are figured out as the most important ingredients for magma tectonic environment discriminations on this dataset.
    2019  .  38(12):    1981-1991    [摘要](53)    [PDF](51)
  • The discrimination between ore-forming and barren granites based on zircon REE compositions: Insights from big data mining
    GENG Ting, ZHOU Yongzhang, LI Xingyuan, WANG Jun, CHEN Chuan, WANG Kunyi and HAN Ziqi
    Yanshanian magmatism is well developed and has obvious metallogenic specificity in Qinzhou-Hangzhou Bays of South China. With the development of in-situ zircon analysis technology, a huge number of zircon composition data has been accumulated in recent years. On the basis of collecting data published by previous researchers, the authors determined the ore-forming potential of rock masses by using zircon REE compositions through big data thinking method, and explored effective geochemical indicators for ore prospecting. Python language was used to program arbitrary combination of elements. A total of 4095 binary diagrams and 121485 ternary diagrams were obtained, and diagrams that could effectively distinguish zircon parent rock metallogenic types were automatically screened out. The results show that different types of ore-forming rocks have different degrees of differentiation. Geochemical indices related to Ce and Eu can be well distinguished, which may result from the oxygen fugacity and water content of magma. Additionally, it is observed that some new element association diagrams (i.e., Dy/Lu-Er/Lu, Gd/Dy-Er/Yb) can distinguish ore-forming types of rock bodies effectively, but the underlying geochemical mechanism has not been fully understood. In brief, the results of geochemical data mining in this paper can be used as the prospecting indicators, which can provide scientific basis for the study and prospecting of Yanshanian hydrothermal deposits in South China, and can also be used to actively explore the application of big data technology in mineralogy.
    2019  .  38(12):    1992-1998    [摘要](77)    [PDF](75)
  • The division of metallogenic prospective areas based on convolutional neural network model: A case study of the Daqiao gold polymetallic deposit
    CAI Huihui, XU Yongyang, LI Zixuan, CAO Haohao, FENG Yaxing, CHEN Siqiong and LI Yongsheng
    Big data and high performance computing make it possible for geology to break through the limitations of various subjective and objective factors and transform from the traditional qualitative description and uncertainty to a more comprehensive quantitative development stage, that is, geology pays more attention to exploring the geological genesis process by mining the correlation between complex and multiple geoscience data. In order to clarify the diversity of geological data in the study area and divide the metallogenic prospective area, the authors aimed to help the geoscientists to make decisions intelligently and efficiently by combining the new methods and technologies of modern informatization. With the Daqiao gold deposit in Gansu Province as the study area, the authors proposed to use one-dimensional convolutional neural network instead of traditional manual calculation and, through training the geochemical and geophysical element data in the study area, excavated the comprehensive metallogenic information in the study area, and then recognized four types of metallogenic prospective areas based on the training results. The results show that the geological mineralization process is complex, and each element of metallogenic prediction plays an important role in the geological mineralization process. On a large scale, the deep learning network model can objectively reflect the nonlinear characteristics of diversified geological data, identify the spatial characteristics of geological elements, extract and excavate the information of mineralization anomalies, and realize the intelligent prediction and evaluation of mineral resources.
    2019  .  38(12):    1999-2009    [摘要](61)    [PDF](60)
  • 3D metallogenic prediction based on machine learning: A case study of the Lala copper deposit in Sichuan Province
    XIANG Jie, CHEN Jianping, XIAO Keyan, LI Shi, ZHANG Zhiping and ZHANG Ye
    Under the background of the vigorous development of big data, the quantitative prediction of mineral resources is the core part of geological big data. The basic idea of comprehensive analysis and mining of multi-information coincides with the concept of big data. With the Lala copper deposit as the study area, the authors carried out 3D mineral resources prediction based on machine learning. In this paper, 3D geological model was established to extract useful information of mineralization and build the quantitative prediction model of the study area. By using the "cube prediction model" prospecting method, the authors adopted the random forest algorithm of machine learning to calculate the probability distribution of mineralization in the study area. In this way, five prospecting prospective areas were delineated. The results show that the random forest has higher prediction accuracy and stability and can make quantitative evaluation on the importance of ore controlling factors. This study has successfully applied machine learning to the 3D mineral resources prediction and made a positive exploration for the prediction and evaluation of mineral resources in the future.
    2019  .  38(12):    2010-2021    [摘要](53)    [PDF](61)
  • Two-dimensional prospecting prediction based on AlexNet network: A case study of sedimentary Mn deposits in Songtao-Huayuan area
    LI Shi, CHEN Jianping, XIANG Jie, ZHANG Zhiping and ZHANG Ye
    There are many challenges in the task of predicting ore deposits from big data repositories. The data are inherently complex and have great significance to the intervenient spatial relevance of deposits. The characteristics of the data make it difficult to use machine learning algorithms for the quantitative prediction of mineral resources. There are considerable interest and value in extracting spatial distribution characteristics from two-dimensional ore-controlling factors'layers under different metallogenic conditions. In this paper, the authors conducted such an analysis by using a Deep Convolutional Neural Network (D-CNN) algorithm named AlexNet. Training on the two-dimensional (2-d) mineral prediction and classification model was performed using data from the Songtao-Huayuan sedimentary manganese deposit. The authors investigated the coupling correlation between the spatial distribution of manganese element, sedimentary facies, outcrop of Datangpo Formation, faults, water system and the areas where manganese orebodies are present, as well as the correlation between different ore-controlling factors by employing the AlexNet networks. After training, the deep convolutional neural network classification model with the verification accuracy of 88.89%, recall of 66.67% and loss value of 0.08 could be obtained. By applying this model to unknown areas for two-dimensional metallogenic prediction, four metallogenic prospective areas. i.e., No. 91, No. 96, No. 154 and No. 184, were delineated, in which the ore potential probability of No. 91 regional ore-bearing probability and No. 154 prospective area is 1, and that of No. 96 is 0.5, suggesting that the probability of existence of undiscovered deposits in prediction areas is large.
    2019  .  38(12):    2022-2032    [摘要](48)    [PDF](52)
  • Construction and prediction of a prospecting model based on recurrent neural network
    ZHANG Yaguang, CHEN Jianping, JIA Zhijie, LI Shi, LIU Suqing, ZHANG Zhiping and ZHANG Ye
    Under the background of big data and artificial intelligence and on the basis of the establishment and application basis of existing traditional geological prospecting model, this paper proposes a prospecting model construction and prediction method based on cyclic neural network, with the purpose of achieving in-depth analysis and understanding of geological data. According to the requirements for construction and prediction of geological prospecting model, the authors combined the data cleaning theory to systematically summarize and summarize the traditional geological prospecting model, thus establishing a geological prospecting knowledge base and providing training data for deep learning algorithms. The accuracy of the comparison results and the time used for classification were comprehensively analyzed. Finally, the RNN classification algorithm was selected to classify the conceptual model of prospecting. In the process of establishing the prospecting model of the study area, by using the key words and ore control elements to complete the model matching, the model was used to analyze the model matching results so as to realize the construction of the regional geological prospecting model and the prediction and analysis of the mineral resources. With the Dashui gold deposit as an example, the construction of the prospecting model was realized quickly and accurately, which effectively provides guidance for the prediction of mineral resources and verifies the feasibility of the method.
    2019  .  38(12):    2033-2042    [摘要](50)    [PDF](55)
  • One-dimensional to three-dimensional density distribution functions and their applications in visualized big data analysis: Exemplified by picritic basalt and some other rocks
    GE Can, ZHANG Qi, LI Xiuyu, SUN He, GU Hai'ou, LI Weiwei and YUAN Feng
    In this paper, the calculation methods and visualization schemes of density distribution functions of different dimensions are proposed to solve the problem of difficulties in analysis and comparison of rock sample data with different orders of magnitude and different measurement errors. Data mining based on the GEOROC and PETDB databases by using the three-dimensional density distribution function of SiO2, total alkali and MgO index as well as the t-distribution random neighborhood embedding visualization method revealed that picritic basalt is similar to oceanite and ankaramite, while picrate is similar to intrusive olivine gabbro and ferropicrate. Comparisons between two-dimensional density distribution function and cumulative density contour visualization were used to analyze the data distribution of different rocks on TAS and Si-Mg maps and the core area of data concentration. It is found that the SiO2 content of magnesium-rich picrite is higher than that of picrite basalt in general distribution. The core area of picrite is mainly located in the B area of TAS diagram, which is contrary to the traditional view that SiO2=45% is used as the boundary between basic and ultramafic rocks.
    2019  .  38(12):    2043-2052    [摘要](49)    [PDF](53)
  • Rock thin section image recognition and classification based on VGG model
    BAI Lin, WEI Xin, LIU Yu, WU Chongyang and CHEN Lihui
    The complexity and multiple solutions of rock thin section images lead to the difficulty in classification of rock thin sections. This paper attempts to apply the deep learning method to the classification of rock thin images. Thin section images of 6 common rock types, such as andesite, dolomite and granite, were selected in the experiment, and 1000 images of each type were used as experimental data. The VGG model was established, and the identification accuracy of the verification set reached 82% after 90,000 iterations. Based on the analysis of the experimental data, the authors found that the rock images with similar compositions are easy to be confused; for example, dolomite and oolitic limestone are both carbonate rocks and it is easy to misjudge each other. Plagioclase porphyry, microcrystalline and cryptocrystalline or vitreous matrix were extracted from the andesite characteristic diagram, and oolitic and interstitial materials were extracted from the oolitic limestone characteristic diagram. The result obtained by the authors proves that the VGG model is effective in the classification of rock thin section.
    2019  .  38(12):    2053-2058    [摘要](55)    [PDF](55)
  • Common mineral intelligent recognition based on improved InceptionV3
    PENG Weihang, BAI Lin, SHANG Shiwei, TANG Xiaojie and ZHANG Zheyuan
    To study 16 kinds of common minerals, the authors collected 1000 images for each type, and then divided them into training set, validation set and test set. Before putting the images into the model, the authors selected a random area of each image for data augmentation. After training the InceptionV3 model with 70000 steps, the authors obtained an 81% accuracy in the test set. Through improving the loss function and introducing the Center Loss, the authors raised the accuracy to 86% after training 400000 steps. The obfuscation matrix shows that, the recognition accuracies for the minerals with obvious appearance characteristics such as malachite are higher while those for other minerals like sphalerite are less due to the obfuscation with other minerals. The analysis of the feature map shows that the model extracts the radial feature of malachite perfectly, and the feature vector of mineral image aggregate is in a high degree, which also can prove the reliability of the model.
    2019  .  38(12):    2059-2066    [摘要](49)    [PDF](44)
  • Research on geological map compilation technology based on spatial data and geological knowledge
    WANG Yanggang, HAO Lirong, HUANG Hui, LI Yusong, ZHANG Dake, ZHANG Qinghe, LI Li, ZHANG Lin and JIHANG Zuorui
    In order to improve the efficiency of specific compilation of geological map and reduce human workload, it is necessary to develop a new method to update the previous geological maps compiled decades ago. After studying the information technology such as AI & BigData, the authors constructed the spatial-temporal model of generalization and the intelligent model of the geological annotation. Then a new technology was developed with which small scale geological maps can be converted to general scale geological maps in a big data environment through compilation. The technology for mapping is called iMapower. Practically, the geological features, such as sedimentary rocks, intrusive rocks, volcanic rocks, metamorphic rocks and structures, can be processed intelligently, interactively, automatically with iMapower based on geological knowledge. The processing procedures include geological body merging, geoline simplification, fault generalization, feature representation, legends illustration, etc. In its application to several pilot areas such as eastern Qinghai Province, Luoyang area, Zhengzhou area and Jingjinji area, the efficiency is obviously improved. The iMapower is proved to be scientific and effective.
    2019  .  38(12):    2067-2076    [摘要](55)    [PDF](56)
  • Geological anomaly extraction based on neighborhood constraint clustering
    YANG Zhaoying, FENG Lei, JIANG Decai, ZHU Yueqin and YU Xianchuan
    Geochemical anomalies often have a strong correlation with ore deposits. The study of effective methods for extracting geochemical anomalies is of great significance for prospecting. The advent of the era of big data and artificial intelligence poses new challenges for the extraction of geochemical anomalies that are automatic and independent of expert knowledge. Geostatistical research shows that it is a new geochemical prospecting idea to identify geochemical anomalies by identifying the special spatial forms of geochemical anomalies, such as lattices, bands, and rings. By analyzing the element value attribute and spatial position of geochemical data, this paper proposes a method based on neighborhood constrained clustering. After clustering geochemical elements, it can extract special shapes such as rectangle, ring and semi-ring and extract geochemical anomalies. In this paper, the geochemical data of two experimental areas in the Xiaoshan area of Henan Province were selected for experiments. The results of Experiment 1 show that the position of the rectangle appears consistent with the location of the known tungsten ore site, whereas the results of Experiment 2 show that the position of the ring is consistent with the location of the known copper orebody. The experiment proves the effectiveness of the method based on neighborhood constrained clustering in extracting geochemical anomalies.
    2019  .  38(12):    2077-2084    [摘要](43)    [PDF](50)
  • 2019  .  38(12):    2085-2102    [摘要](74)    [PDF](63)

Information Bulletin


Journal Retrieval

Quoted rankings


Online journals