The exposition of minerals to oxygen as well as non-treated tailings in mining activities alter the balance of the ecosystem, specifically leading to the generation of acidic solutions in the presence of sulfidic minerals. Several secondary iron minerals are precipitated in these settings that can be detected via remote sensing applications. The purpose of this research is to investigate the capacity of hyperspectral analysis to determine the abundance of Acid Mine Drainage (AMD)-indicator secondary iron minerals in mine sites with the guidance of ground truth information. To this end, we focus on an abandoned coal mine site in Turkey to detect secondary iron minerals associated with AMD via multispectral Sentinel-2 imagery, in accordance with the laboratory analysis of field-collected samples through X-Ray Diffraction (XRD), Inductive Coupled Plasma (ICP), and ASD spectral analysis. In relation with the conducted laboratory XRD and ICP results, the proposed methodology first reveals the iron-induced absorption feature located between 700 and 900 nm on field-collected ASD spectra and reference USGS spectra through a baseline method, namely parabola fitting method. The subsequent remote sensing analysis then applies hyperspectral unmixing to Sentinel-2 imagery and identifies the spectral endmember indicating iron-absorption behavior by computing its spectral angle distance to reference spectra. The experiments reveal that while the iron absorption characteristics are not apparent in pixel spectra, the utilized unmixing methodology enables capturing of those features at sub-pixel level on the resulting endmembers. Second, the comparison between the calculated abundances with unmixing and iron levels obtained with ground based ICP analysis indicate coherent correlation values. Finally, among the utilized unmixing methods, the performance of SISAL is found better than MVSA with the resulting correlation values of 0.76 and 0.63, respectively, while also returning closer endmembers to the reference iron spectra. The performed research demonstrates the potential of hyperspectral applications on Sentinel-2 data to uncover the sub-pixel iron-induced spectral features in the visible region, proving compatible results between the spectral and laboratory analysis.

在采矿活动中矿物暴露于氧气以及未处理的尾矿会改变生态系统的平衡，特别是在存在硫化矿物的情况下导致产生酸性溶液。在这些环境中会沉淀出一些次生铁矿物质，可以通过遥感应用程序进行检测。这项研究的目的是在地面真相信息的指导下，研究高光谱分析的能力，以确定矿山中酸性矿山排水（AMD）指标的二次铁矿物的丰度。为此，根据对通过X射线衍射（XRD）对现场采集的样本进行的实验室分析，我们将重点放在土耳其的一个废弃煤矿现场，以通过多光谱Sentinel-2图像检测与AMD相关的二次铁矿物质，电感耦合等离子体（ICP）和ASD光谱分析。对于进行的实验室XRD和ICP结果，所提出的方法首先通过基线方法（即抛物线拟合法）揭示了场收集的ASD光谱和USGS参考光谱上700-900 nm之间铁诱导的吸收特征。然后，后续的遥感分析将高光谱分解应用于Sentinel-2图像，并通过计算其与参考光谱的光谱角距离来识别表明铁吸收行为的光谱末端成员。实验表明，尽管铁吸收特性在像素光谱中不明显，但利用的混合方法可以在最终的端部件上以亚像素级捕获这些特征。第二，通过基于地面的ICP分析获得的解混后的丰度与铁含量之间的比较表明了相关的相关值。最后，在使用的解混方法中，发现SISAL的性能优于MVSA，其相关值分别为0.76和0.63，同时还使更接近的末端成员返回到参考铁谱。进行的研究表明，在Sentinel-2数据上应用高光谱的潜力可揭示可见区域中亚像素铁诱导的光谱特征，从而证明了光谱分析与实验室分析之间的兼容结果。发现SISAL的性能优于MVSA，其相关值分别为0.76和0.63，同时还使更近的端基返回参考铁谱。进行的研究表明，在Sentinel-2数据上应用高光谱的潜力可揭示可见区域中亚像素铁诱导的光谱特征，从而证明了光谱分析与实验室分析之间的兼容结果。发现SISAL的性能优于MVSA，其相关值分别为0.76和0.63，同时还使更近的端基返回参考铁谱。进行的研究表明，在Sentinel-2数据上应用高光谱的潜力可揭示可见区域中亚像素铁诱导的光谱特征，从而证明了光谱分析与实验室分析之间的兼容结果。