Spectral Mixture Modeling with Laboratory Near-Infrared Data I: Insights into Compositional Analysis of Europa

Abstract

Europa's surface composition and physical characteristics are commonly constrained using spectral deconvolution through linear mixture (LM) modeling and radiative transfer-based (RT) intimate mixture modeling. Here, I compared the results of these two spectral modeling- LM versus RT- against laboratory spectra of water (H$_{2}$O) ice and sulfuric acid octahydrate (SAO; H$_{2}$SO$_{4}$$\cdot$8H$_{2}$O) mixtures measured at near-infrared wavelengths ($\sim$1.2-2.5 $\mu$m) with grain sizes of 90-106 $\mu$m (Hayes and Li, 2025). The modeled abundances indicate that the RT more closely reproduces the laboratory abundances, with deviations within $\pm$5% for both H$_{2}$O ice and H$_{2}$SO$_{4}$$\cdot$8H$_{2}$O with $\sim$100 $\mu$m grains. In contrast, the LM shows slightly larger discrepancies, typically ranging from $\pm$5-15% from the true abundances. Interestingly, both LM and RT tend to consistently overestimate the abundance of H$_{2}$SO$_{4}$$\cdot$8H$_{2}$O and underestimate H$_{2}$O ice across all mixtures. Nonetheless, when H$_{2}$SO$_{4}$$\cdot$8H$_{2}$O either dominates (>80% as observed on Europa's trailing hemisphere; Carlson et al. 2005) or is present only in trace amounts ($\sim$10% on areas in Europa's leading hemisphere; Dalton III et al. 2013; Ligier et al. 2016), both the LM and RT render acceptable results within $\pm$10% uncertainty. Thus, spectral modeling using the RT is preferred for constraining the surface composition across Europa, although the LM remains viable in specific compositional regimes.