Spectral-library matching

Spectral-library matching answers the question: what material is present at this location, based on the absorption features in its reflectance spectrum? The method compares an observed spectrum against a reference library of known material signatures to identify the best-matching composition, focusing on diagnostic absorption features rather than overall spectral shape.

The Tetracorder algorithm (Clark et al. 2003) is the foundational implementation.^[1] Its pipeline applies continuum removal first, dividing the observed spectrum by its convex-hull continuum to isolate absorption features from the baseline. Each continuum-removed spectrum is then compared against library entries in spectral regions defined by diagnostic rule-sets for each material group. Match quality is scored as the coefficient of determination (r^2) of the continuum-normalized comparison. Additional quality gates screen for minimum absorption feature depth, acceptable reflectance level, and continuum slope consistency. The best-matching material group is identified from the highest-quality passing match.

NASA JPL applied Tetracorder to EMIT imaging spectrometer data from the International Space Station to produce the EMIT L2B mineral identification product, identifying 10 mineral groups including carbonates, clays, sulfates, and iron oxides at 60m resolution.^[2] The product (EMITL2BMIN v001) covers land surfaces globally at ISS orbit swath since 2022.^[3]

The method requires an imaging spectrometer with continuous spectral coverage across diagnostic absorption regions (typically 400-2500 nm VNIR-SWIR) and sufficient spectral resolution (under 10 nm sampling) to resolve absorption features. It fails on spectrally featureless surfaces, under cloud cover, with high aerosol loading that distorts continuum determination, and for materials absent from the reference library.