Atmospheric correction method and application of satellite hyperspectral data: A case study in mineral resource monitoring

Hyperspectral sensors have become indispensable tools in remote sensing applications, playing a pivotal role in precision agriculture, mineral alteration mapping, land cover classification, and autonomous satellite wildfire detection. Their high spectral resolution and comprehensive spectral information about ground objects underscore their significance. However, despite the wealth of hyperspectral data sources, challenges in data processing have impeded their quantitative application on the Earth's surface. Atmospheric correction, a critical procedure for deriving surface reflectance that accurately captures surface properties, is essential to overcoming these challenges. In this context, we present an adaptive and automated atmospheric correction scheme specifically designed for hyperspectral data. This approach involves the inversion of atmospheric parameters, including aerosol optical depth (AOD) and precipitable water vapor content (PWV), utilizing the inherent hyperspectral information in the image. Importantly, the atmospheric correction is performed without the need for simultaneous atmospheric and surface observations. AOD retrievals exhibit excellent consistency with ground measurements (RMSE < 0.05, R2 > 0.78), and PWV retrievals are also accurate, with an RMSE less than 0.17 g/cm2 and an R2 greater than 0.94. The acquired surface reflectance demonstrates a remarkable resemblance in terms of shape (spectral angle < 2.2°) and magnitude (RMSE < 0.02) compared to in-situ measurements. The ZY-1 02D Satellite (ZY02D), led by the Ministry of Natural Resources of China, is purposefully designed for monitoring natural resources. The Advanced HyperSpectral Imager (AHSI) aboard ZY02D covers a wide area of 60 km * 60 km with a medium-to-high resolution of 30 m and a spectral range spanning from 400 to 2500 nm, featuring 166 channels. To monitor mineral resources in Guangdong Province, China, ZY02D AHSI images are acquired. After atmospheric correction using our method, the resulting surface reflectance spectral curve is utilized to identify mineral areas. The similarity between satellite-based and in-situ measured surface reflectance is assessed by spectral angle and Euclidean distance to identify potential mineral areas. The absorption characteristics of minerals are extracted from the satellite-based surface reflectance to enhance the results. Through comparison with mining rights maps, more than 80% of mining area maps are successfully identified. Moreover, it facilitates the monitoring of unauthorized mining activities, thereby improving the enforcement efficiency of government agencies.