地下矿藏寻踪记：高光谱成像如何让岩石「开口说话」？

Tracing Hidden Mineral Deposits: How Hyperspectral Imaging Makes Rocks "Speak"

在甘肃柳园的戈壁深处，一支地质勘探队正面临着挑战：如何在绵延数百平方公里的裸露岩层中，精准锁定铜矿的蛛丝马迹？十年前，这样的任务需要地质队员背着地质锤和罗盘，顶着烈日徒步数月，用肉眼辨识岩石颜色的微妙差异。而今天，他们只需放飞一架搭载高光谱成像仪的无人机——随着镜头扫过苍茫大地，岩石的“光谱指纹"在屏幕上跃动，隐藏的铜矿脉络如同被施了魔法般浮现。

这双穿透地表的“光谱之眼"，正是高光谱成像技术。它不仅能看清人眼无法捕捉的矿物光谱特征，甚至能通过岩石的细微反光差异，破解地球亿万年的地质密码。

Deep in the Gobi Desert of Liuyuan, Gansu, a geological exploration team faces an unprecedented challenge: how can they accurately pinpoint traces of copper ore across hundreds of square kilometers of exposed rock formations? A decade ago, such a task would have required geologists to carry rock hammers and compasses, trekking for months under the scorching sun, relying on the naked eye to detect subtle variations in rock coloration. Today, they simply launch a drone equipped with a hyperspectral imager. As the sensor scans the vast terrain, the "spectral fingerprints" of the rocks flicker across the screen, and hidden copper veins emerge as if by magic.

This "spectral eye" that penetrates the Earth's surface is hyperspectral imaging technology. It not only captures mineral spectral characteristics invisible to the human eye but also deciphers billions of years of geological history through subtle differences in rock reflectance.

对现场采集的36个样品（左一）在实验室进行了高光谱和传统方法的分析和表征。根据常见的地球化学数据对样品进行聚类，并自动识别样本（左二）。光谱图利用聚类材料的反射率建立（右一）

A total of 36 field-collected samples (left) were analyzed and characterized in the laboratory using both hyperspectral and traditional methods. Samples were clustered based on common geochemical data, with automated sample identification performed (second from left). Spectral graphs were constructed using the reflectance of clustered materials (right).  

高光谱成像仪的核心能力，源于它惊人的“视觉灵敏度"。以爱博能提供的HY-1710系列为例，其光谱范围覆盖400-2500nm，相当于同时捕捉可见光、近红外和短波红外的共460个波段。这就像给相机装上了460层滤光片，每一层都能捕获特定矿物的独特“光波密码"——比如高岭石在2200nm附近的吸收峰。

The core capability of hyperspectral imaging lies in its remarkable "visual sensitivity." Take the HY-1710 series as an example: its spectral range covers 400–2500 nm, capturing 460 bands across visible, near-infrared, and short-wave infrared wavelengths. It is as if the camera is equipped with 460 layers of filters, each capable of detecting the unique "light-wave signature" of specific minerals—such as the absorption peak of kaolinite near 2200 nm.

矽卡岩矿床中常见蚀变矿物光谱曲线 / Spectral Curves of Common Alteration Minerals in Skarn Deposits

在甘肃华牛山多金属矿区，某团队使用高光谱成像仪对裸露岩层进行遥感检测，旨在寻找新的矿藏资源。通过高光谱技术，他们对岩石进行分类和映射，在广阔的区域内快速识别出铅矿、锌矿等矿物。此外，团队运用了高效的图像处理方法，提升了矿数据信息的提取精度，尤其是在铅锌矿体方面获得了81%的矿物识别率。这使得他们成功绘制出矿物分类图，并识别出了该地区的重要矿藏。

华牛山矿区经历了多次地质演变，形成了复杂的构造特征和多样的矿藏类型，通过高光谱技术，团队不仅为资源的探索提供了科学依据，也为今后的矿藏开发和地质研究奠定了坚实的基础。这种“从地表透视地壳运动"的能力，让高光谱成像成为新一轮找矿突破战略行动的核心技术。

In the Huaniushan polymetallic mining area of Gansu, a team used hyperspectral imaging to conduct remote sensing detection of exposed rock formations, aiming to discover new mineral resources. Through hyperspectral technology, they classified and mapped rocks, rapidly identifying minerals such as lead and zinc ores across extensive areas. Moreover, the team employed efficient image processing methods to enhance the accuracy of mineral data extraction, achieving an 81% mineral recognition rate for lead-zinc ore bodies. This enabled them to successfully generate mineral classification maps and identify significant deposits in the region.

The Huaniushan mining area has undergone multiple phases of geological evolution, resulting in complex structural features and diverse mineral deposit types. By applying hyperspectral technology, the team not only provided a scientific basis for resource exploration but also laid a solid foundation for future mineral development and geological research. This ability to "see through surface features to interpret crustal movements" has made hyperspectral imaging a core technology in the new strategic action for mineral exploration breakthroughs.

研究团队绘制的华牛山矿物分布图 / Mineral Distribution Map of Huaniushan Produced by the Research Team

2024年，中国地质调查局西安矿产中心联合航空物探遥感中心和资源所，共同申报的基于国产高光谱数据找矿远景区快速圈定技术，已成功入选中国地质调查局正式印发的《新一轮找矿突破战略行动先进适用勘查技术清单（第一批）》。更值得一提的是，我国高分五号卫星的高光谱载荷，已实现90%以上的矿物识别准确率，该团队更获得了2024年度中国科学院杰出科学成就奖。

In 2024, the Xi’an Mineral Resources Center under the China Geological Survey, in collaboration with the Aerogeophysical Survey and Remote Sensing Center and the Institute of Mineral Resources, successfully developed a rapid prospect delineation technology based on domestically produced hyperspectral data. This achievement was included in the *List of Advanced and Applicable Exploration Technologies for the New Strategic Action for Mineral Exploration Breakthroughs (First Batch)* officially issued by the China Geological Survey. It is also worth highlighting that the hyperspectral payload onboard China’s Gaofen-5 satellite has achieved over 90% accuracy in mineral identification. The team behind this effort was awarded the 2024 Outstanding Scientific Achievement Prize by the Chinese Academy of Sciences.

由西安矿产中心发布的东天山-北山成矿带高光谱遥感矿物分布图

Hyperspectral Remote Sensing Mineral Distribution Map of the East Tianshan-Beishan Metallogenic Belt (Released by the Xi’an Mineral Resources Center)

从戈壁荒漠到深海矿藏，高光谱成像正在重写人类探索地球的方式。它不仅是技术的革新，更是一种认知的革命——当我们学会倾听岩石的光谱语言，那些沉睡亿万年的矿藏终将揭开面纱，为文明续写新的能源篇章。

From desert wilderness to deep-sea mineral resources, hyperspectral imaging is reshaping how humanity explores the Earth. It is not only a technological revolution but also a cognitive one—by learning to interpret the spectral language of rocks, we are unveiling mineral deposits that have slumbered for billions of years, paving the way for a new chapter in energy resources and civilizational progress.

案例来源 / Sources：

1. 叶彬,王加昇,李金龙,付浩.高光谱遥感技术在地质勘查中的应用现状及展望[J].化工矿物与加工,2024,53(11):77-88

2. Wan, Yq., Fan, Yh. & Jin, Ms. Application of hyperspectral remote sensing for supplementary investigation of polymetallic deposits in Huaniushan ore region, northwestern China. Sci Rep 11, 440 (2021)

3. 李士杰，何海洋，秦昊洋，孙旭，王思琪，刘小玉. 中国地质调查局西安矿产资源调查中心. 东天山-北山成矿带高光谱矿物填图数据集. 2024,09

4. 基于国产高光谱数据找矿远景区快速圈定技术入选《新一轮找矿突破战略行动先进适用勘查技术清单（第一批）》

5. 董新丰,甘甫平,李娜,闫柏琨,张磊,赵佳琪,于峻川,刘镕源,马燕妮.2020.高分五号高光谱影像矿物精细识别.遥感学报,24(4): 454-464