Capability assessment of Sentinel-2 imagery for graphite deposits exploration

Abstract

Graphite, a critical raw material for high-technology industries, is used as an anode in lithium-ion batteries due to its high thermal and electrical conductivity and high-temperature resistance. The study evaluates the use of Sentinel-2 satellite images in exploring new graphite deposits, demonstrating the effectiveness of remote sensing methods and comparing different classification methods. It also introduces new graphite indices, the Normalized Difference Graphite Index (NDGI) and the Graphite Band Math Index (GBMI), which differentiate the spectral signatures of graphite mineralization from other land cover spectral signatures. The study focuses on creating a reference graphite mineralization spectral library, developing an image-based optimum graphite spectral library from Sentinel-2 satellite images, evaluating the performance of Spectral Angle Mapper (SAM), Spectral Information Divergence (SID), and Constrained Energy Minimization (CEM) classification methods, and selecting suitable threshold values for classification. The Kütahya-Oysu Graphite Mine in Turkey was chosen as the training area, while other test sites include Xinghe-Huangtuyao Graphite Mine (Inner Mongolia, China), the Pingdu Graphite Deposit (Shandong, China), the Jixi-Liumao Graphite Deposit (Heilongjiang, China), the Luobei-Yunshan Graphite Deposit (Heilongjiang, China), the Balama Graphite Project (Mozambique), and the Molo Graphite Project (Madagascar). In Oysu, a significant graphite deposit in Turkey, graphite ore is formed in metamorphic graphite schist and graphite-bearing muscovite schist. The Oysu's graphite ore is microcrystalline and macrocrystalline type of graphite and displays unique absorption at 704 and 2225 nm of full-spectrum wavelengths, as well as at 740 or 783 nm of the spectral detection range of Sentinel-2. The Sentinel-2 sensor, which has been widely used in geological applications, produces useful data for graphite exploration on a large scale. Sentinel-2's high spectral resolution in the VNIR region makes it useful for exploring graphite deposits. Data-driven approaches like NDGI and GBMIs indices reveal graphite-related mineralization areas, while supervised classification methods map graphite ratios using reference end-member spectral. A knowledge-based probabilistic algorithm, SID, measures spectral discrepancy probability, resulting in more successful graphite mineralization mapping. Lower threshold values (0.050–0.060) increase the probability of pure graphite areas.