[A Method to Extract Content of Minerals Based on Measured Hyperspectral Data]

Area of Science:

• Geoscience
• Spectroscopy
• Mineralogy

Context:

• Accurate mineral content extraction is crucial for geological surveys and resource exploration.
• Traditional linear decomposition models face challenges with spectral data complexity and noise.
• Existing spectral matching methods can be limited by spectral library completeness and noise interference.

Purpose:

• To develop an advanced method for mineral content extraction from rock spectra (350-2500 nm).
• To enhance the accuracy of mineral identification and quantification using spectral matching and the Hapke model.
• To improve upon linear decomposition models by incorporating mineral symbiotic relationships and sectional noise filtering.

Summary:

• A novel method utilizes rock spectra (350-2500 nm), spectral matching, and the Hapke model to transform reflectance into single scattering albedo for mineral content determination.
• The method incorporates sectional noise filtering and regional mineral spectra libraries to boost identification accuracy.
• Validation on the Baogutu rock body showed high accuracies for identifying quartz (75%), feldspar class (100%), and altered minerals (92.2%), and for extracting content of feldspar class (80.5%), hornblende (64%), and altered minerals (92.36%).

Impact:

• The method theoretically improves accuracy by using single scattering albedo, avoiding complex nonlinear calculations.
• Incorporating mineral symbiotic relationships enhances the reliability of mineral identification.
• The approach offers significant guidance for rapid analysis of alteration information in geological studies.