Translational Imaging Spectroscopy for Proximal Sensing

Area of Science:

• Geosciences
• Spectroscopy
• Remote Sensing

Background:

• Proximal sensing, the near-field counterpart of remote sensing, has diverse applications.
• Translational laboratory imaging spectroscopy is valuable for various research topics.
• Geoscientific applications necessitate precise hyperspectral data preprocessing for at-surface reflectance retrieval.

Purpose of the Study:

• To introduce GeoMAP-Trans, a novel preprocessing chain for at-surface reflectance retrieval in hyperspectral data.
• To adapt and evaluate this chain for the HySPEX VNIR/SWIR imaging spectrometer system.
• To assess the performance and accuracy of the proposed algorithm for geological mineral samples.

Main Methods:

• Developed a preprocessing chain (GeoMAP-Trans) with radiometric, geometric, and spectral modules.
• Adapted the chain for the HySPEX VNIR/SWIR imaging spectrometer.
• Incorporated the Reduction of Miscalibration Effects (ROME) framework for radiometric accuracy.
• Evaluated performance using standard image quality metrics and comparative spectrometer measurements.

Main Results:

• GeoMAP-Trans achieved high qualitative results with broad applicability.
• High radiometric accuracy was achieved using the ROME framework.
• Geometric accuracy exceeded 1 μpixel.
• Spectral accuracy was better than 0.02% for the full spectrum and >98% for absorption features.

Conclusions:

• The GeoMAP-Trans algorithm provides accurate at-surface reflectance retrieval for hyperspectral data.
• The generic design offers broad applicability in geoscientific research.
• Empirical evidence shows differences between point and imaging spectrometers for non-Lambertian samples.