金矿床遥感影像模式及证据权重法在找矿预测中的应用  

APPLICATION OF REMOTE SENSING IMAGERY MODEL FOR GOLD DEPOSIT AND "EVIDENCE-RIGHT-WEIGHT"METHOD IN GOLD DEPOSIT PREDICTION

作　　者：罗富生[1] 崔振奎[1] 徐双 王国娟 

机构地区：[1]核工业北京地质研究院

出　　处：《国土资源遥感》1994年第4期23-29,7,共8页Remote Sensing for Land & Resources

摘　　要：河南熊耳山地区为我国金矿的重要成矿区，为加速金矿资源评价和找矿勘探工作，利用遥感技术对该区金矿成矿规律进行了研究，并进行了金矿找矿靶区优选。首先利用多种方法进行ＴＭ数字图像处理，以获取理想的遥感图像，然后进行遥感影像信息解译分析，结合区域地质调查和典型金矿床解剖研究，确立主要控矿因素（如构造、含矿地层、岩体、蚀变等）。根据金矿分布规律和从遥感图像上提取的综合影像特征，建立了不同金矿床类型遥感影像模式。在此基础上应用证据权重图形信息复合方法进行找矿靶区优选。该方法以贝叶斯法则为基础，将有利的控矿地质因素表示为控矿地质证据的二值指示图形，求出它们的权重。利用地理信息系统将对矿床相对独立的地质因素，又称矿化证据，进行统计图形复合，以求出矿床分布的后验概率，根据矿化后验概率的分布来指示矿化远景区。该方法属于统计方法在地学应用中的新进展，方法本身是由遥感技术、统计方法、人工智能结合而来的，既有理论上的严密性又易于在地质领域应用，并具有直观、快速、灵活的特点。实践表明，根据不同金矿化类型遥感影像模式和证据权重图形信息复合方法进行找矿靶区优选是科学的，也是成功的。在预测的８片Ａ类和５片Ｂ类远景区中，经野外验证，均不同程度?Abstract Xiongershan region, Henan province is an important gold-mineralized area in China. Research of gold mineralization and selection of prospective targets in the region have been carried out by using remote sensing technology. First of all, TM image data are processed through various methods to get ideal remote sensing images, then, the image information is interpreted and analyzed in combinaticon with regional geological survey and typical gold deposit study to determin main ore control factors, such as structures,ore-bearing strata, rock bodies, and alteration, etc. The remote Sensing imagery models of different types of gold deposit have been set up on tho basis of extraction of comprehensive image signatures and gold deposit distribution pattern.In the light of the models, prospecting targets have been chosen by using 'evidence-right-weight' graphics information integration. The method,which presents favourable geological factors of ore control as binary graphics and calculates their 'evidence-right-weight', is based on Bayes rule, By use of geographic information system (GIS), mutual independent geological factors relative to the deposit, mineralization evidence, are integrated statisticaly for calculating posterior probability of gold deposit distribution,which can be used to indicate potential gold-mineralized area, The method,combination of remote sensing, statistics and expert's experiences, is provided with a rigorous theoretical basis and visual, fast, and flexible characterastics,and is convinent for geological application.After practice, it is concluded that according to remote sensing imagery models of different type of gold mineralization and 'evidence-right-weight'graphics information integration method to choose prospecting targets is succeesful. Various extents of gold mineralization have been found in 8 first class and 5 second class potential areas predicted by us after field testing . The result is very obvious.

关 键 词：金矿床 遥感 影像模式 权重法 找矿 

分 类 号：P618.510.8[天文地球—矿床学]