水系沉积物低弱地球化学异常提取方法研究

发布时间：2018-03-23 16:40

本文选题：水系沉积物　切入点：低弱地球化学异常　出处：《吉林大学》2017年博士论文

【摘要】：我国于上世纪70年代开展的1:20万水系沉积物地球化学测量工作已基本覆盖山区和半山区。1:20万水系沉积物地球化学测量是低成本、高效率的矿产勘查方法,在圈定成矿远景区(带)、寻找矿田和大、中型矿床,特别是有色、贵金属矿床的勘查中发挥了极其关键的作用。随着我国矿产勘查工作的不断深入,地表矿、浅部矿以及易识别矿逐渐减少,找矿的重点逐步转向寻找隐伏矿、半隐伏矿和难识别矿。目前我国积累的海量高质量的1:20万水系沉积物地球化学数据仍将在未来的地质找矿中发挥重要作用。经过多年的工作,虽然1:20万水系沉积物的“高、大、全”地球化学异常几乎被评价殆尽,但其中仍蕴含的大量有找矿价值的低、弱地球化学异常信息值得进一步挖掘。研究表明,大矿不一定有大异常,大异常不一定有大矿,矿致异常的规模与强度受诸多因素制约。近些年我国在低弱异常区陆续发现一系列矿床的事实充分证明了低弱地球化学异常的找矿价值。目前,尚无完善的水系沉积物低弱地球化学异常识别的有效方法,因此研究探索低弱地球化学异常识别理论与方法具有实际意义。本文以湖南省1:20万水系沉积物为主要研究对象,同时选择了其他景观区的水系沉积物进行了相关研究。论文重点研究了湖南省水系沉积物的地球化学特征,剖析了水系沉积物元素地球化学背景的影响因素,研究了地球化学背景影响因素的消除方法,在此基础上,提出了基于正态分布的异常提取方法和基于多元非正态总体的异常提取方法。湖南省水系沉积物地球化学成分具有明显的空间差异,通过K均值聚类方法将湖南省分为六个地球化学区,不同地球化学区的元素背景值具有一定的差异。研究表明,水系沉积物元素地球化学背景受研究尺度、岩性背景、样品粒度、景观条件等多种因素的影响,其中基岩岩性的影响尤为显著。在岩性复杂区,由于水系沉积物来源于多个“母体”,元素含量概率分布通常呈现出多“母体”的混合分布,元素含量概率分布图上往往表现出“多峰”的特点。地质背景(特别是岩性背景)对元素背景的影响,其结果可能导致有价值的低弱地球化学异常被掩盖,同时可能误将元素背景含量高的地质体圈定成“异常”。为了消除岩性背景的影响,本文提出了以K均值聚类和稳健回归分析为核心的地球化学背景影响因素消除方法。该方法以水系沉积物中可以表征岩性特征的造岩元素以及亲石元素含量为基础,通过K均值聚类算法将样品分为多个子类,检查每个子类中元素与主要氧化物(如Si O2、Fe2O3)的相关性。对于与主要氧化物无明显相关关系的元素,首先对该元素数据进行正态变换,剔除异点后计算均值和标准差,用均值和标准差对数据进行标准化。对于与主要氧化物呈显著相关关系的元素,建立该元素与主要氧化物的回归方程,采用稳健回归方法求解待定系数,计算回归值和标准残差,然后用回归值和标准残差对数据进行标准化。最后形成标准化的正态分布的基础数据。以湖南省桂阳地区和浏阳地区1:20万水系沉积物数据为例进行了分析,发现原始数据多为偏态或多峰分布,而经过处理的数据均接近均值为0、标准差为1的标准正态分布,在很大程度上消除了地质背景的影响。消除背景影响的标准化数据突出了一些低弱地球化学异常,压制了高背景,使得异常与已知矿床的吻合更好。对于样品粒级、系统误差以及景观条件的影响,本文采用了分幅标准化方法进行消除。本文分析了常规等值线异常图的弱点,提出了基于正态分布的标准化数据元素组合异常色块编图法,采用主成矿元素与成矿伴生元素相结合的方式,将采样点位以一个色块表示,每个色块分为9个小方格,每个方格代表一种元素,然后采用相应倍数的标准偏差表示异常强度。研究表明,该方法可以传达更丰富的关于异常强度以及矿床剥蚀程度的信息,可以提高异常圈定的准确度。该方法充分考虑了热液矿床元素垂直分带规律和研究区不同矿种异常元素组合特点。与传统等值线法相比具有信息量大、异常位置准确、易使用等特点。以消除背景影响的标准化正态数据为基础,对传统的马氏距离背景与异常划分方法进行了改进,提出了基于多元正态总体的异常提取方法和马氏距离异常图编制方法。本文提出的马氏距离异常图具有简单、直观的特点,可以展示出每个点位的异常元素组合信息。通过实例研究,发现马氏距离方法圈定的异常位置与已知矿床吻合较好,突出了弱小异常。利用多维数据信息,从多元角度刻画元素异常,有利于进行低弱异常识别,提高异常的预测精度。研究表明,水系沉积物中一些元素的含量,即使消除了岩性背景的影响,其标准化数据也难以符合正态分布,甚至近似正态分布。在这种情况下,采用基于正态分布理论的水系沉积物地球化学异常提取方法是不合适的。为此,本文提出了适合非正态分布高维数据的基于实码加速遗传算法的投影寻踪模型,将高维数据投影到一维子空间,依据投影值的大小来提取地球化学异常。该方法不仅可以解决上述方法的数据正态分布限制,也可以充分利用矿化过程中亏损元素的信息,充分利用数据中的有效信息并避免其他信息的干扰。实例研究表明,采用消除背景的标准化数据的投影寻踪方法可以发现传统方法无法发现的低弱异常信息。水系沉积物低弱地球化学异常提取是地球化学数据处理过程中的关键问题,本文提出的一系列方法对于背景消除理论方法方面有了一定的创新和补充。基于消除背景的标准化数据的多元地球化学异常提取方法能够有效地识别低弱异常,压制高背景,将多元数据融合的化探异常评判也符合多元信息集成的发展趋势。本文提出的方法对于低弱地球化学异常识别与提取方面具有一定的理论以及实际意义。
[Abstract]:1:20 000 stream sediment geochemical survey in China in the last century in 70s has basically covered.1:20 of mountainous and semi mountainous 000 stream sediment geochemical survey is of low cost, high efficiency method in mineral exploration, delineated metallogeneticprospective provinces (zones), looking for ore field and large, medium-sized deposits, especially nonferrous, plays the key role of the exploration of precious metal deposits in mineral exploration in China. With the deepening of the work, surface mine, shallow mine and easy to identify ore prospecting focus gradually reduced, and gradually turn to search for concealed deposits, half concealed ore deposits and ore difficult to identify. Play an important role in massive accumulation of China's high quality at 1:20 000 stream sediment geochemical data will remain in the future geological prospecting. After years of work, although the 1:20 000 stream sediment "high, big, the whole" geochemical anomaly evaluation was almost exhausted, but the Still contains a large number of a prospecting value low, weak geochemical anomaly information mining. Further research shows that large deposits do not necessarily have unusual large abnormalities does not necessarily have large deposits, ore size and strength of the anomaly is restricted by many factors. In recent years our country in the low anomaly areas have found a series of ore deposits the fact proves that the weak geochemical anomaly prospecting value. At present, there is no perfect effective method to identify the low sediment geochemical anomaly, so exploring weak geochemical anomaly identification theory and method has practical significance. This paper takes Hunan province 1:20 000 stream sediment is the main research object, and choose the other sediments the landscape was studied. The paper focuses on the geochemical characteristics of stream sediment in Hunan Province, analyzes the elements of stream sediment geochemical background The influence factors on the method of eliminating the effect of geochemical background factors, on this basis, put forward the anomaly extraction method based on normal distribution and based on the multiple abnormal extraction method. The overall Hunan province stream sediment geochemical composition has obvious differences in space, through the K means clustering method divides Hunan province six geochemical regions, with some differences in background values in different geochemical zones. The study shows that the elements in stream sediment geochemical background of scale, lithological background, sample size, a variety of factors which affect the landscape conditions, bedrock lithology is particularly significant. In areas with complex lithology, due to sediment source in the "matrix", mixed distribution elements in the probability distribution are usually more "parent", elements of probability distribution often show multi peaks" The characteristics of geological background (especially lithologic background) influence on the element background, could result in lacking of valuable geochemical anomalies to be covered, and may mistakenly delineate geological body element background with high content of "abnormal". In order to eliminate the influence of lithology background, using K means clustering and robust the regression analysis method to eliminate the influence factors of geochemical background as the core. The method to build rock sediments can characterize lithologic characteristics of elements and lithophile element content as the basis, the sample is divided into several sub classes by K means clustering algorithm, the elements of each sub class were examined with major oxides (such as Si O2, Fe2O3) for the correlation. And there was no obvious correlation between the main oxide elements, first normal transformations of the data elements, get rid of abnormal after the calculation of the mean and standard deviation of data using the mean and standard deviation Standard. For there was a significant correlation with the main oxide elements, establish regression equation of the main elements and oxides, using robust regression for the undetermined coefficient method, regression calculation value and standard error, and then use the return value and standard error of standard data. Finally the formation of basic data distribution of normal standard by 1:20. 000 stream sediment data in Hunan area and Liuyang area of Guiyang Province as an example to analyze, and find that the original data is more skewed or multi peak distribution, and the processed data are close to the mean value is 0, the standard deviation is 1 of the standard normal distribution to a great extent to eliminate the effect of geological background. Standard data. Eliminate the background effect highlights some of the weak geochemical anomalies, suppress the high background, the anomaly and agree better known deposits. For sample size, system error and landscape The effects of the condition in the picture of the standard method eliminated. This paper analyses the conventional contour map of abnormal weakness, put forward the normal distribution of the standard data element mapping method based on the combination of abnormal color, the main ore-forming elements associated with mineralization elements combination, the sampling points to a color, each color is divided into 9 small squares, each square represents an element, then the standard deviation of the corresponding multiples of said anomaly intensity. The results show that the method can convey more about abnormal strength and deposit erosion degree information, can improve the accuracy of the anomaly. This method takes full account of the hydrothermal deposit elements of vertical zoning and research area of different mineral anomaly elements. Compared with the traditional contour method with a large amount of information, the abnormal position is accurate, easy to use features to eliminate. Standard background of normal data based on Mahalanobis distance background and anomaly classification method to improve traditional proposed anomaly map compilation method of anomaly extraction method and Mahalanobis distance in multivariate normal population. Based on Mahalanobis distance anomaly map is proposed in this paper has a simple, intuitive, can show the abnormal elements combining the information of each point. Through case study, found that the delineation of Mahalanobis distance method and abnormal position of known deposits in good agreement, highlighting the weak anomalies. Using multidimensional data, from the multi angle description element abnormal, to identify the weak anomalies, improve the prediction accuracy of the anomaly. The study shows that the content of some elements in stream sediments the even eliminates the influence of lithology background, the standard data to conform to normal distribution, and the approximate normal distribution. In this case, using the base Yu Zheng normal distribution theory of stream sediment geochemical anomaly extraction method is not appropriate. Therefore, this paper put forward a non projection pursuit model of real coded accelerating genetic algorithm based on the normal distribution of high dimensional data, high-dimensional data is projected to one dimensional subspace, based on the projection values to extract the geochemical anomalies. The method can not only solve the above method data of normal distribution, also can make full use of the loss element mineralization in the process of information, make full use of effective information interference in the data and avoid other information. The case study shows that the elimination of projection pursuit method standard data background can be found in the weak abnormal information can not be found by traditional methods low. Stream sediment geochemical anomaly extraction is the key problem in geochemical data processing, a series of methods proposed for back On the view of innovation and elimination theory method is supplemented. Multiple Geochemical Standard data to eliminate background anomaly extraction method can effectively identify the weak anomaly based on the suppression of high background, the development trend of the multivariate data fusion geochemical anomaly evaluation with multi information integration. This method has a certain the theoretical and practical significance for the identification of weak geochemical anomalies and extraction.

【学位授予单位】：吉林大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：P632

【相似文献】