Effective delineation of rare metal-bearing granites from remote sensing data using machine learning methods: A case study from the Umm Naggat Area, Central Eastern Desert, Egypt
Effective delineation of rare metal-bearing granites from remote sensing data using machine learning methods: A case study from the Umm Naggat Area, Central Eastern Desert, Egypt
复制标题
DOI：
10.1016/j.oregeorev.2022.105184
复制
发表时间：
2022-11-02
影响因子：
3.3
通讯作者：
Csamer, Arpad
中科院分区：
文献类型：
--
作者：
Abdelkader, Mohamed A.;Watanabe, Yasushi;Csamer, Arpad
钠长花岗岩（ABG）被认为是稀有金属（RMs）最重要的宿主之一。因此，通过适当的岩性判别，充分识别ABG，可以大大提高稀有金属资源的找矿目标。为了从卫星数据中圈定ABG露头，本研究整合了8种图像增强技术，包括最佳指数因子、假色复合、频带定量、相对频带深度、独立成分分析、主成分分析、去相关拉伸、最小噪声分数变换和光谱指数比，对ASTER和Sentinel-2 （S2）数据集进行了解译。这种综合方法可以有效地识别埃及中东部沙漠乌姆纳格特地区的AGB露头。这些综合图像处理技术得到的解释图在现场进行了系统验证，并为支持向量机算法（SVM）支持的不同岩性特征选择过程（即训练和测试数据圈定）奠定了基础。为了获得高质量的岩性解释图，将支持向量机应用于Sentinel-2、ASTER和ASTER- s2数据集。经我们的实地调查和之前的地质图证实，融合的ASTER-S2分类正确地描绘了ABG。此外，通过综合构造分析（林元提取及其密度图）和热液蚀变探测，验证ABG、高密度带和高蚀变区分布之间的空间关联，从而揭示新的潜在矿化带和建议的勘探目标。研究表明，新发现的ABG主要分布在研究区的南部和西南部，并确认了乌姆纳格特地区北部已知矿化带的位置。ABG的分布及其与蚀变带和高构造密度带的空间相关性表明，研究区内稀有金属成矿主要受构造控制（NW、NNW、NNE和N-S），表明研究区内稀有金属有较高的交代富集可能性。我们的研究基于svm支持的ASTER-S2数据解释，提供了研究区域的最新地质图。重要的是，研究结果揭示了乌姆纳格特地区稀有金属成矿的巨大勘探潜力，为后续地球化学土壤测量工作确定了新的异常。
Albitized granite (ABG) is considered as one of the most significant hosts of rare metals (RMs). Consequently, adequate recognition of ABG through proper lithological discrimination highly increases the targeting of rare metal resources. In order to delineate outcrops of ABG from satellite data, our study integrates eight image enhancement techniques, including optimum index factor, false color composites, band rationing, relative band depth, independent component analysis, principal component analysis, decorrelation stretch, minimum noise fraction transform, and spectral indices ratios, for the interpretation of ASTER and Sentinel-2 (S2) datasets. This integrated approach allows the effective discrimination of AGB outcrops in the Umm Naggat area, Central Eastern Desert, Egypt. The interpretation maps derived from these integrated image processing techniques were systematically verified in the field and formed the base for the feature selection process (i.e., training and testing data delineation) of different lithologies supported by the support vector machine algorithm (SVM). In order produce a high-quality lithological interpretation map, SVM was applied to Sentinel-2, ASTER, and combined ASTER-S2 datasets. The fused ASTER-S2 classification properly delineates ABG, as verified by our field vestigations and confirmed by previous geological maps. Furthermore, comprehensive structural analysis (lin-eaments extraction and their density map) and hydrothermal alteration detection were performed to check the spatial association between the distribution of ABG, higher density zones, and highly altered areas, that in turn, could shed light on new potentially mineralized zones and proposed exploration targets. Our study reveals new ABG occurrences mainly situated in the southern and southwestern parts of the study area, and it confirms the location of known mineralized zones in the northern part of the Umm Naggat region. The distribution of ABG and its spatial correlation with alteration and high structural density zones suggest that rare-metal mineralization mostly structurally controlled (NW, NNW, NNE, and N-S), demonstrating the higher possibility of metasomatic enrichment of rare-metals within the study area. Our study provides an updated geological map of the study area based on the SVM-supported interpretation of ASTER-S2 data. Importantly, the results reveal a high exploration potential for rare-metal mineralization at Umm Naggat and defining new anomalies for follow-up work geochemical soil surveys.