資源探査に関する研究の動向と大規模データプロセッシング

[3601-07-07] Effectiveness of an integration of hyper- and multi-spectral images with geological information for accurate mapping of metal deposit-related minerals

○Nguyen Tien Hoang¹, Katsuaki Koike¹, Shuho Noda², Tetsuichi Takeda², Ayumi Kosugi², Kazuo Masuda², Koki Kashiwaya¹, Yusuke Ota¹ (1. Kyoto University, 2. JOGMEC)

司会：木崎 彰久（秋田大学）

Keywords:Geologic remote sensing, Mineral identification, ground truth, HyMap image, WorldView-3 image

Multispectral and hyperspectral remote sensing have been widely used for mineral mapping and resource exploration. Hyperspectral imagery has proven its capability for detailed discrimination of various types of Earth surface materials because of having hundreds of bands with high spectral resolution. Despite using a small number of bands, multispectral imagery with high spatial resolution is also effective in terms of signal to noise ratio and identifying mineral groups with small distribution areas. This research sheds new light on data combination of hyperspectral and multispectral remote sensing, geology, and field survey for improving the accuracy of mineral identification. For this, an integration of HyMap, a hyperspectral sensor covering the 0.4–2.5 micro-m range with 126 contiguous spectral bands, and WorldView-3, the first multi-payload and highest resolution commercial satellite sensor, was selected to test the accuracy of mineral mapping in the Cuprite area in the western United States. The study area consists of two main centers, which were formed by extensive development of the late-Miocene advanced argillic zones of iron-bearing minerals, clays, micas, sulfates and carbonates. More than 100 rock samples were collected during a field survey for the ground truth data, and were then measured their spectral reflectance characteristics by FieldSpec 4 spectroradiometer and their mineral compositions were identified by X-ray diffraction. HyMap image was processed using the Mixture Tuned Matched Filtering (MTMF) technique integrated with several mineral indices calculated from WorldView-3 image for producing the final mineral map. The result proves the key roles of hyperspectral remote sensing and of multiple data integration method for mineral exploration.