The Primary Uncertainties in Infrared Spectral Studies of Mars

Introduction: A key goal for infrared remote sensing studies of Mars is to determine which minerals are present and absent. Visible/infrared spectroscopy is the primary method for remote mineral identification. In order to understand the resulting interpretations, it is necessary to understand the sources of uncertainty. Here we detail the primary uncertainties in the method. Interpretations of remote sensing data sets of Mars currently differ widely. For example, different researchers interpret non-detection of a mineral differently. Some interpret non-detection as absence of the mineral, while others note that minerals in the field commonly exhibit weak or absent spectral bands [1]. The wide ranging compositional interpretations cause changing or contradictory geologic interpretations. Thus the same remote sensing data sets are interpreted as pointing to a “cold/dry Mars” and a “watery past.” The resulting confusion leads researchers to ask, what are the uncertainties in these data sets? Here we illustrate three of the four primary uncertainties. We also explain work needed to address the current contradictions and confusion. The four primary uncertainties are: (1) Limited ability to detect unknown minerals, i.e., when we have no reference spectrum (no sample). (2) Variations in surface texture that cause huge uncertainties in mineral abundance calculations. (3) Misunderstandings of how much mineral is required for detection. This uncertainty impacts interpretations of what non-detection means. (4) Coatings cause significant uncertainties, but due to space, we do not cover them here. Mars analog data: One of our Mars analog sites is Coso Hot Springs, California. In June 2005, we acquired hyperspectral imagery over the full terrestrial optical range. SpecTIR measured airborne hyperspectral images covering 0.4-2.5 μm. SEBASS measured midand thermal infrared (2.5-5 and 7.5-12.5 μm) hyperspectral imagery. RamVan measured ground based, MiniTES analog hyperspectral imagery (7.5-12.5 μm). Uncertainty #1: Unknowns: Minerals are detected by searching for a match to a signature measured of a sample. However, when we lack a matching library signature, i.e., lack a sample (Fig.1), then detection has poorer sensitivity. For Mars, we do not know what samples are missing from the library, and thus what unknown minerals may be present and undetected. Uncertainty #2: Abundance determination: In order to calculate abundance, the spectral contrast of the target is compared to the spectral contrast of a reference spectrum. However, increasing surface roughness and decreasing particle size can decrease the target spectral contrast. Thus we need a reference spectrum that has the same spectral contrast as the target, e.g., that was measured of a material with the same surface roughness and particle size. Fig.2 illustrates how an incorrect reference spectrum leads to an incorrect abundance estimate [2]. The black trace shows a reference laboratory spectrum, measured of 100% gypsum. Current abundance calculations assume that a field deposit composed of 100% gypsum will exhibit the same spectral contrast as the laboratory spectrum. The red and blue traces show field (MiniTES analog) spectra measured of gypsum outcrops, of a rough-surfaced gypsum (red trace), and a smooth-surfaced gypsum (blue trace). Table 1 shows that the calculated abundances are incorrect. That occurs because the reference spectrum did not have the same spectral contrast as the targets, because the materials have different surface textures.