Differentiating Lunar Surface Hydration from Thermodynamic Artifacts: A High-Latitude Analysis of Chandrayaan-2 IIRS Data

<jats:p>The Chandrayaan-2 Imaging Infrared Spectrometer (IIRS) provides high-resolution (80 m) hyperspectral data crucial for mapping lunar surface mineralogy and hydration. The definitive detection of lunar water, specifically through the 3.0 µm OH/H2O absorption feature, requires the rigorous removal of solar-contaminated thermal emission. This thermal glow is modeled and removed using Planck’s Law of blackbody radiation via the Verma et al. (2022) algorithm. In this study, we apply this pipeline to a longitudinal IIRS strip extending from mid-latitudes into the lunar South Pole (∼71◦S), mapping mineralogical abundances via K-Means clustering and continuum-removed band parameters. Most critically, we identify and characterize a fourstage algorithmic cascade in polar cold-trap pixels that produces a synthetic 3.0 µm hydration feature indistinguishable from genuine surface water. The cascade originates in photometricnot thermal—correction: the Lommel-Seeliger formula amplifies long-wavelength sensor noise (up to 5.3× within this orbital strip; theoretical maximum 20× at the µ0 clamp floor) at high incidence angles (i &gt; 75◦), driving apparent reflectance at 5 µm to unphysical values. The Verma algorithm correctly recognizes no valid thermal signal and issues a QA=3 failure flag, but a “do no harm” data-preservation protocol passes the amplified spectrum to the final dataset uncorrected. Subsequent convexhull continuum removal anchors to the artificially elevated 5 µm spectral tail—a region ISRO’s own instrument documentation flags as unreliable due to low detector SNR—synthesizing apparent 3 µm band depths of ∼45% in pixels with no physical absorption whatsoever. Critically, this artifact does not trigger any QA warning: 341,372 QA=0 validated pixels (10.8% of all valid data, ∼2,185 km2) silently carry the cascade signal. We distinguish this artifact from genuine hydration and from DFSAR subsurface ice detections, and propose that future pipelines mask pixels by incidence angle threshold prior to photometric correction.</jats:p>