新疆雪米斯坦铀成矿有利区高光谱遥感识别方法研究  

An Identification Method of Hyperspectral Remote Sensing for the Favorable Uranium Metallogenic Area in Xuemisitan,Xinjiang

作　　者：韩晓青[1] 杨国防[1] 木红旭 杨云汉 HAN Xiaoqing;YANG Guofang;MU Hongxu;YANG Yunhan(Beijing Research Institute of Uranium Geology,Beijing 100029,China)

机构地区：[1]核工业北京地质研究院,北京100029

出　　处：《世界核地质科学》2022年第2期309-320,共12页World Nuclear Geoscience

基　　金：国防基础科研计划稳定支持专题“甘肃龙首山鈾成矿区航空高光谱花岗岩智能识别与分类技术研究”;中国铀业2022年地勘项目“铀矿勘查高光谱遥感填图技术研究及试验应用”联合资助

摘　　要：新疆雪米斯坦铀成矿区是我国重要的火山岩型铀成矿区,区内产出我国境内首次发现的火山岩型铀-铍矿床,具有良好的综合找矿前景。为了在雪米斯坦火山岩地区进一步寻找铀成矿有利区,推动雪米斯坦铀成矿区实现找矿突破,在前人基础上,对研究区铀成矿要素进行了分析,认为区内构造、侵入岩、岩石类型、蚀变等为铀成矿有利要素。其中研究区断裂构造十分发育,按照尺度分为一、二、三、四级,所有断裂构造属性被详细统计;侵入岩主要为华力西晚期各类侵入岩,其中以花岗岩最为发育;研究区与成矿有关的蚀变矿物种类十分丰富,主要蚀变矿物为:高铝绢云母、中铝绢云母、低铝绢云母、绿泥石、高岭石、碳酸盐、蒙脱石等。为了实现铀成矿有利区准确预测,首先采用深度学习方法对区内机载SASI高光谱数据进行蚀变矿物提取,确定高铝绢云母、中铝绢云母、低铝绢云母、绿泥石、高岭石和碳酸盐等矿物中心位置和矿物分布面积;其次对区内构造、侵入岩、岩性组合和蚀变矿物等铀成矿要素进行空间分析,分别获得该区断裂密度分布图、侵入岩影响范围图、岩石组合分类图和蚀变矿物分类缓冲图;最后采用证据权法将成矿要素空间分析结果作为证据因子,计算了证据权值,并进行成矿后验概率计算,最终根据全部计算结果圈定5个成矿远景预测区,为今后雪米斯坦铀矿区进一步找矿提供了遥感技术支持。Xinjiang Xuemistan area is an important volcanic rock type uranium metallogenic area in China and produced the first volcanic rock type uranium-beryllium deposit in China,indicating good comprehensive prospecting prospect.In order to find more favorable areas for uranium mineralization in this area,and promote the breakthrough of uranium exploration in this area,we studied the uranium metallogenic factors based on the predecessors and find that the structure,intrusive body,volcanic rock type,alteration are controlling factors for uranium mineralization.The fault in the study area are well developed which were divided into 4 grade according to the scale,the attributes of the 4 grade fault structures were studied in detail.Alteration minerals related to mineralization were extracted and the main altered minerals are high-aluminum sericite,medium aluminum sericite,low-aluminum sericite,chlorite,kaolinite,carbonate,montmorillonite,etc.In order to achieve accurate prediction of favorable uranium mineralization areas,deep learning method was firstly used to extract altered minerals from the airborne SASI hyperspectral data.,and The anomaly center and area of alteration minerals such as kaolinite and carbonate were determined.The spatial relation analysis was conducted for the uranium metallogenic elements such as structure,intrusive rocks,lithological assemblages and altered minerals to create the maps of fracture density distribution,intrusive rock influence range,lithological assemblages and alteration mineral classification.The weight evidence method was used to calculated the metallogenic evidence factors and the metallogenic posterior probability by using the spatial analysis results.According to The calculation results,five metallogenic prospective prediction areas were delineated,which provided remote sensing technical support for the future.uranium prospecting in Schemistan uranium mining area.

关 键 词：雪米斯坦火山岩 SASI高光谱数据 蚀变矿物深度学习提取 铀成矿有利区预测 

分 类 号：P619.14[天文地球—矿床学] P627[天文地球—地质学]