Testing Mineral Specificity for Mapping with ZY1-02E and ASTER at the Murzuq–Tibesti Margin, Southern Libya

Mineral maps can overstate identification when detectable absorptions remain compatible with several minerals. We present a framework that tests candidate labels for a sensor and scene before assigning them pixel by pixel, retaining family evidence where subclasses remain unresolved. We applied it to ZY1-02E AHSI observations of southern Libya, using background-adjusted matched-subspace, wavelet and shape measures for SWIR absorptions and the complete VNIR feature for ferric iron. Thresholds derived from scene backgrounds were evaluated on held-out blocks using numerical mixtures and independent mineral confusers. The broad 2.30 µm response was readily recovered but accepted 36.2% of tested Al-OH mixtures and was withheld. Conditional calcite and Fe/Mg-OH branches had cross-assignment of at most 1.33%. Among accepted Al-OH pixels, 31.2% remained at family level. Release stability varied: Al-OH passed in four of 13 background-sampling designs, whereas the broad 2.30 µm group failed throughout. ASTER and mapped geology were used to contextualise the evidence and identify eight potential follow-up zones covering 34.5 km2, with differing persistence under selection and registration changes. The framework links controlled spectral separability to the level of detail retained in mineral maps. Independent observations are still required to confirm mineral identity and geological origin.