Effect of halite coatings on thermal infrared spectra

Abstract

Characterizing the occurrence and distribution of soluble salts on planetary surfaces allows us to model or monitor aqueous and geochemical conditions. Thermal infrared (TIR) remote sensing is useful for identifying some types of salt deposits; however, mineral abundance determinations are based on the interaction of infrared with only the top few hundred micrometers of the surface. Thus, distinguishing massive deposits from coatings presents a challenge to TIR remote sensing. To better understand the TIR properties of spectrally transmissive halite coatings, we investigated the effects of coating thickness and texture on the TIR reflectance spectra of halite-coated glasses. We analyzed two coating textures with variable thickness: (1) continuous and (2) discontinuous particulate coatings. As halite coating thickness increases, the intensity of substrate absorption bands decreases nonlinearly. The substrate is not detected with halite coatings >150 μm thick. Therefore, when coatings are present, it is not possible to apply TIR models of mineral abundances using linear deconvolution algorithms. We show that continuous and coarse particulate halite coatings increase the reflectance minimum (emissivity maximum), making them potentially detectable on planetary surfaces by the methods of Osterloo et al. (2008). However, the reflectance minimum does not change if coatings have low areal coverage (<50%) or are composed of fine particles (<5 μm). Thus, halite that forms as efflorescent coatings or has been redistributed as fine dust is not detected using TIR spectroscopy. Our results explain why Cl salts are not detected with TIR spectroscopy at many locations on Mars despite high Cl contents.

Constraints on the composition and particle size of chloride salt‐bearing deposits on Mars

“…Another possibility is that additional evaporite phases do occur in these regions but are (1) present at low enough abundances (likely less than a few percent) to remain undetected in both VNIR and MIR data or (2) buried by the chloride salt deposits and undetectable. Recent laboratory work has shown that thick coatings of halite can obscure the spectral signatures of materials underneath at MIR wavelengths [ Berger et al , ], although this has yet to be verified at VNIR wavelengths.…”

Fitting the curve in Excel®: Systematic curve fitting of laboratory and remotely sensed planetary spectra

“…The facsimile spectrum is overlain atop the observed spectrum of an unknown sample or surface, and modified by either varying amounts of individual contributors, e.g., a 0.25 + b 0.5 + c 0.25 = x, and or adding or subtracting more contributors, e.g., a 0.25 + b 0.25 + c 0.15 + d 0.1 + e 0.05 +f 0.05 + g 0.05 + h 0.05 + i 0.03 + j 0.02 = x, until an acceptable match with the observed spectrum is produced which appears analogous to the person performing the curve matching or until some threshold residual for the facsimile versus observed spectrum is met (e.g., McSween et al, 2003, and references therein). As a first pass, the technique can be illustrative if one has very little context for the spectrum under investigation but it requires a spectral library and algorithms (e.g., Ramsey and Christensen, 1998) to produce and fit the deconvolved spectrum/facsimile to some residual or human input to suggest or rule out contributors, and it does not capture the fact that spectra of mineral assemblages rarely, if ever, mix linearly (Singer, 1981;Clark, 1999;Kraft et al, 2003;Berger et al, 2015). As a technique it may also ignore other factors that can contribute to altering a particular mineral, or an assemblage of mineral spectra, such as, grain size, porosity, texture or roughness, grain packing, temperature, phase angle, atmosphere, adsorbed species, spatial resolution, spectral resolution, distance, the probable effects of space weathering, and so on.…”

Recurring slope lineae and chlorides on the surface of Mars

“…A linear deconvolution is therefore often used to assess the quantitative abundance of minerals using TIR remote sensing [Thomson and Salisbury, 1993;Ramsey, 1996;Ramsey and Christensen, 1998;Feely and Christensen, 1999;Hamilton and Christensen, 2000;Christensen et al, 2004;Huang et al, 2013]. However, chloride minerals behave nonlinearly in the TIR as a result of spectral transmission/low maximum emissivities and a lack of spectral features; therefore, linear deconvolution is an unreliable method of assessing their areal abundance [Eastes, 1989;Osterloo et al, 2008;Berger et al, 2015]. The linear deconvolution technique was not used in this study because an accurate positive detection of chlorides cannot be made given the areal abundance estimate from linear deconvolution results.…”

“…Therefore, the densest regions of large RSL provide the highest potential for positive detection of a chloride lag deposit. The detection limit of chloride coatings in the thermal infrared was explored by Berger et al [2015] by studying a range of optically thin (<150 μm) thicknesses. Thermal infrared reflectance minima (emission maxima) did not change significantly between optically thick (1 mm) and the thinnest continuous chloride coating in their study (29 ± 2 μm) [Berger et al, 2015].…”

“…The detection limit of chloride coatings in the thermal infrared was explored by Berger et al [2015] by studying a range of optically thin (<150 μm) thicknesses. Thermal infrared reflectance minima (emission maxima) did not change significantly between optically thick (1 mm) and the thinnest continuous chloride coating in their study (29 ± 2 μm) [Berger et al, 2015]. Optically thick coatings caused an increase in reflection minima (decrease in emission maxima) by 10% reflectance, where all optically thin coatings increased reflection minima by~5%.…”