Planetary processes active and ancient: Hollowing on Mercury, ancient crust formation on Mars, and identifying Mars-analog habitats.

This dissertation comprises a thermophysical model that shows elemental sulfur may be involved in the potentially active processes that form enigmatic features called hollows on the Mercurian surface, a suite of remote sensing techniques that unveil anorthositic rocks in ancient martian crust, and deep learning to discover the spatial resolutions necessary to identify astrobiology targets in images of Mars analog landscapes.

On Mercury, hollows are high-reflectance, flat-floored depressions observed nearly globally. Hollows are thought to form via sublimation, or a “sublimation-like” process, but the identity of the sublimating phase is poorly constrained. To better understand which phase might be responsible for hollow formation, I calculated sublimation rates for 57 candidate hollow-forming volatile phases at Mercury-relevant conditions. I found that elemental sulfur is the phase most likely responsible for hollow formation. Several limitations with previously published hollow-formation models lead me to suggest a novel model, inspired by terrestrial thermokarst processes, that I dub “Sublimation Cycling Around Fumarole Systems”.

Mars offers a unique opportunity to constrain models of planetary differentiation given its observable record of ancient crust and its middling size among terrestrial planets. Unlike on Earth, where plate tectonic processes have erased the earliest rocks from the geologic record, remnants of martian crust older than ~4.1 Ga (pre-Noachian) are exposed at the surface. Here I present evidence for an extensive, pre-Noachian layered igneous complex north of Hellas basin containing feldspathic rocks, which I interpret as anorthositic. These feldspathic outcrops raise the intriguing possibility that the Hellas-forming impact uplifted a sample of a deep, global, low-density component of the ancient martian crust.

With sample return as an imminent goal for the NASA Mars Exploration Program, precise targeting of rocks and outcrops with significant science value, especially those with the highest probability of containing biosignatures, is a top priority. The increased focus of scientific objectives on locating biosignature-bearing outcrops comes with a need for more precise identification of rover-explorable targets from orbit. Therefore, I established a deep-learning based method for determining the spatial resolutions necessary to identify high-priority, astrobiology targets with a specified confidence level and applied this method to the identification of habitats in images of Mars-analog landscapes. This study serves as a template for quantifying identification confidence as a function of image spatial resolution for habitats across many Mars-analog environments, which could inform the decision making-process for future targets of Mars exploration.

This is a revised drafted that incorporates revisions from the graduate school.