Identification and Extraction of Silica Mine in Sanbaidun, Yumen City

Hyperspectral technology, while imaging the spatial features of a target object, disperses each spatial pixel to form dozens or even hundreds of narrow bands for continuous spectral coverage. The resulting spectral data can be visualized using an image cube, where two dimensions represent space and the other represents the spectrum. In this three-dimensional space integrating spectral and spatial information, "continuous" spectra and diagnostic feature spectra of ground features can be arbitrarily obtained. This allows for direct identification of target features based on spectral knowledge and further quantitative information about the features. In geological applications, mineral identification and information processing technologies can be categorized into three main types: characteristic parameters of single diagnostic absorption, complete waveform features, and spectral knowledge models.
Hyperspectral mineral mapping background
Since the launch of the US EO-1 satellite in November 2000, its Hyerion hyperspectral imager, with a spectral range of 400–2500 nm, 220 bands, spectral resolution of 10 nm, geometric resolution of 30 m, and bottom imaging width of 7.5 km, has been used as a geological remote sensing data source. However, its narrow imaging width makes it difficult to apply over a large area, which limits its application and promotion. my country's GF-5 satellite (launched in May 2018) carries a visible-shortwave infrared hyperspectral camera with a spectral range of 400–2500 nm, a spectral resolution of 5 nm for the VNIR channel and 10 nm for the SWIR channel, a geometric resolution of 30 m, and a swath width of 60 km.
Current applications demonstrate that hyperspectral imaging of surface rock alteration information is highly effective. It can identify specific alteration types and even single mineral species, such as limonite, biotite, potassium feldspar, chlorite, epidote, and sericite, representing a significant breakthrough for remote sensing mineral exploration. With continuously improving spectral resolution, hyperspectral remote sensing mineral identification is gradually evolving from identifying mineral types to recognizing finer information such as mineral subclasses and mineral composition. Furthermore, as practical applications deepen, the demand for detailed mineral information is increasing. However, spectral resolution and mineral identification methods are the main factors limiting the fine-grained identification of hyperspectral minerals.

Zhongke Xiguang, after in-depth market research, has developed a mineral monitoring satellite for the mining industry. This satellite carries both short-wave infrared and panchromatic payloads. The short-wave infrared payload utilizes the 2000-2500nm spectral range, where mineral spectra are most concentrated, achieving a spectral resolution of 15nm, a swath width better than 12km, and a spatial resolution of 20 meters. When fused with the panchromatic payload, the spatial resolution reaches 5 meters. The satellite has a designed weight of 120kg. It is currently in the overall design phase and is scheduled for launch in February 2025.
Yumen City (Research Area)
Yumen City, located in western Gansu Province, was formerly known as Yangguan. It is an important node city on the Silk Road and a significant petrochemical base in my country. In recent years, Yumen City has actively responded to the national "Belt and Road" initiative and the "Western Development" strategy, vigorously developing the new materials industry and achieving remarkable results.

△ Figure 1 Study Area
Sanbaidun Silica Mine, Yumen City
Silica is a non-metallic mineral primarily composed of silica-rich minerals. Silica deposits include quartz sandstone, quartzite, quartz sand, vein quartz, and metasomatic siliceous hornfels. The main minerals are quartz and chalcedony. The chemical industry requires quartzite and vein quartz to have a silica content ≥90%, with lower water absorption and porosity being preferable, ideally zero. All are mined via open-pit mining.
The Sanbaidun silica mine is located in Yumen City, Gansu Province, approximately 44 km from the city center. Administratively, it falls under the jurisdiction of Huangzhawan Township. Its central geographical coordinates are: East longitude: 97°20′03″; North latitude: 40°37′11″.
The following is the spectral reflectance of the Sanbaidun silica mine obtained from the hyperspectral image of XIGUANG-003 satellite after data preprocessing, atmospheric correction and spectral reconstruction. The results show that the spectrum of XIGUANG-003 satellite is in good agreement with the ASTER JPL spectral library, thus verifying the correctness of the XIGUANG-003 satellite spectrum.

△ Figure 2 Reflectance map of Baitun silica mine (XIGUANG-003 star)

△ Figure 3 Reflectance map of Baitun silica mine (ASTER JPL spectral library)

△ Figure 4 Sanbaidun Silica Mine, Yumen City

