This item is under embargo and not available online per the author's request. For access information, please visit


Date of Award

Spring 5-15-2019

Author's School

Graduate School of Arts and Sciences

Author's Department

Earth & Planetary Sciences

Degree Name

Doctor of Philosophy (PhD)

Degree Type



We are currently living in the golden age of robotic exploration of Mars, with a continued robotic presence there since 1997. Next to Earth, Mars is the planet about which we have gathered the most geologic information. Unlike Earth, Mars does not appear to have plate tectonics, and the planet’s primary and secondary crust is dominated by basalts. Understanding the compositional diversity of the materials that make up the martian crust will give us a better insight into the geologic processes that formed the planet and its subsequent evolution. One large and growing source of martian surface compositions is the Alpha Particle X-Ray Spectrometer, an in-situ instrument that has been carried on three Mars rovers, Spirit, Opportunity, and Curiosity. This instrument has measured elemental compositions for martian rocks and soils across three separate terrains on Mars. This dissertation seeks to characterize the diversity and quantify the similarities of compositions of rocks and rock suites as reported in the APXS datasets as well as published compositions of the martian meteorites. The careful application of multivariate statistics allows for a rigorous assessment of these diverse compositions to explore possible compositional groupings and primary and secondary relationships. To this end, a statistical grouping model comprised of hierarchical clustering and a similarity index, informed by image analysis and ground-truth in-situ Mars exploration, is applied to the data.

Verification of models is essential as statistics can provide spurious results. In Chapter 2, I apply the statistical grouping model to a set of well characterized Opportunity APXS data. These datasets have significant covariant information, such as geographic relationships, rock textures, geologic context, Pancam spectra, and established working hypotheses about local geology. Chapter 2 also explores how the statistical model works relative to compositions of rocks that have undergone different surface treatment by the rock abrasion tool (RAT) on Opportunity. I also test the model sensitivity to dust/soil contamination. The model is able to reproduce several well-known relationships among the Endeavour crater and Meridiani Planum lithologies, as well as produce some new geologic interpretations. New interpretations include: The Meridiani Plains Burns formation is compositionally diverse enough to parse into two superclusters, mostly along the lines of surface coatings. Analysis of the data by excluding S and Cl, some Burns formation rock compositions are similar to the Endeavour crater Shoemaker impact breccias. The clastic Grasberg formation is compositionally homogenous across two temporally distinct units and is most similar to the Shoemaker impact breccia in the Endeavour crater rim segments. These relationships support a local erosional origin instead of a distal ash origin. The lowest member of the Shoemaker breccia, the Copper Cliff unit, is compositionally similar to the pre-Endeavour Matijevic formation and contains Matijevic-type spherules, indicating that this unit contains eroded Matijevic materials. The Matijevic formation is compositionally distinct from other Endeavour materials but is similar to the “blue” (in Pancam false color) basaltic rocks Marquette Island and Margaret Brush. As a final example, the new regolith breccia class of martian meteorites (NWA 7034/7475) is the only class of martian meteorite to represent common martian surface compositions in the APXS data sets.

Chapter 3 expands the analysis to data collected by Opportunity within Perseverance Valley, using the statistical grouping model to classify lithologies and compare them to rock suites examined elsewhere along the Endeavour crater rim. The model establishes four rock lithologies within the valley, making it the most lithologically diverse location since the rover’s first exploration of Endeavour crater at Cape York. The lithologies include: a clast-poor impact breccia that forms the walls of the valley, an outcrop of resistant basaltic rocks that appear “blue” in false color Pancam imagery, an outcrop of pitted rocks that are some of the most silica-rich materials examined by Opportunity, and the valley floor material that comprises a loose regolith mixture of impact breccia, Meridiani soil, and “blue” rocks, implying a trough filled with locally mass-wasted materials. The “blue” rocks are similar enough in composition and texture to be classified as members of the lithology of “blue” rocks observed on the rim overlooking Marathon Valley, an outcrop that is also co-located with a pitted, silica-rich rock unit. This similarity, combined with the lateral offset of units across the valley, indicates that Perseverance Valley is a graben, formed along a radial impact fault that lowered the “blue” and pitted rocks ~80 meters to their current position. Evidence for aqueous alteration and modern aeolian erosion rounds out the valley’s history, and a formation model for Perseverance Valley, as supported by observations and lithologic relationships exposed by the statistical grouping model, is presented.

The fourth and final chapter of this dissertation expands the statistical grouping model to include representative lithologies from all three of the landings sites where an APXS was deployed, Meridiani Planum and Endeavour crater (Opportunity), Gusev crater (Spirit), and Gale crater (Curiosity). This study also includes a set of representative martian meteorite compositions to best establish what these data sets, APXS compositions and published martian meteorite compositions, can tell us about the lithochemical diversity of the martian crust. The basalts of Gusev crater divide into three discrete classes along the lines of geography, pointing to limited mixing across units. In general, Gusev plains basalts are most similar to Endeavour crater materials, and they share a distinctive Fe/Mn mantle compositional trend compared to that of the Shergottite basalts. Unlike Gusev rocks, the sedimentary rocks of Gale crater mostly do not cluster along geographic lines, but where they do, they appear to be related by a common sediment source (Bathurst and Bell Island). Gale crater has two primary igneous protoliths that are especially alkaline in composition (Jake M and Clinton). All three rover locations include relatively silica-rich lithologies, which have some of the most dissimilar compositions across all of the data sets. Some of these silica enrichments are likely due to secondary processes (Endeavour pitted rocks, and Greenhorn Gale class), but for those that may be indicative of primary silica enrichments (ALH 84001, Clinton Gale erratic, and Buckskin Gale class) their SiO2 vs FeO/MgO ratios could be indicative of a magma evolutionary trend, potentially that of a hydrous, calc-alkaline source. Considering compositions from all of the data sets, the Endeavour crater Grasberg formation composition is still more similar to the average Shoemaker formation composition than to any of the other rock suites represented in the analyses. The “blue” rocks of Endeavour crater can be subdivided into different rock suites, and the “blue” rock suite that is interpreted here to be pre-Endeavour impact (including the Perseverance Valley and Marathon Valley overlook “blue” rocks) is similar enough to the pre-Endeavour clastic Matijevic formation that it is possible the Matijevic formation represents a lithified sedimentary rock formed from eroded pre-Endeavour “blue” rocks. The Meridiani plains erratic, Marquette Island, is also a member of this pre-Endeavour “blue” rock class. Owing to its position on top of the Burns formation, Marquette Island must have been excavated from a younger, larger crater, such as Iazu or Bopolu, indicating the pre-Endeavour “blue” suite of rocks may be continuous across tens of kilometers and could represent the Noachian cratered terrain underling the Burns formation of Meridiani Planum, some of the oldest rocks examined in-situ on Mars.

The application of this specifically tuned statistical grouping model provides new insights into the geochemical relationships between rock suites within, and across sites on Mars. It provides a rapid and quantitative tool for assessing large numbers of targets at once, and can help to provide a better understanding of the geologic units and context of the landing sites, as well as comparing and contextualizing the APXS data set with martian meteorites to better understand the diversity of the martian crust.


English (en)

Chair and Committee

Bradley L. Jolliff

Committee Members

Raymond E. Arvidson, Robert F. Dymek, William B. McKinnon, Scott M. McLennan,


Permanent URL:

Available for download on Sunday, April 18, 2021