Massive open data resources are changing the way that people do science. To make use of those data resources, data science methods and technology can be leveraged by stakeholders of various disciplines. The objective of this paper is to present our experience of using visual exploratory data analysis as a method to facilitate collaboration and hypothesis generation in geoscience research. The research team consisted of both geoscientists and computer scientists. A use case-driven, iterative approach was applied to create a collaborative and communicative environment. Through several rounds of use case analysis and technological development, a data visualization pilot system was created for studying the co-relationships between chemical elements and mineral species. The exploratory data analyses conducted in those use case studies led to several research hypotheses for future work. This research illustrates the usefulness of exploratory data analysis for hypothesis generation in a data science process. Although the presented project is in geoscience, the discussed method and experience can also be translated into other disciplines.

The Mars Science Laboratory rover, Curiosity, is using a comprehensive scientific payload to explore rocks and soils in Gale crater, Mars. Recent investigations of the Bagnold Dune Field provided the first in situ assessment of an active dune on Mars. The Chemistry and Mineralogy (CheMin) X-ray diffraction instrument on Curiosity performed quantitative mineralogical analyses of the <150m size fraction of the Namib dune at a location called Gobabeb. Gobabeb is dominated by basaltic minerals. Plagioclase, Fo56 olivine, and two Ca-Mg-Fe pyroxenes account for the majority of crystalline phases along with minor magnetite, quartz, hematite, and anhydrite. In addition to the crystalline phases, a minimum similar to 42wt % of the Gobabeb sample is X-ray amorphous. Mineralogical analysis of the Gobabeb data set provides insights into the origin(s) and geologic history of the dune material and offers an important opportunity for ground truth of orbital observations. CheMin's analysis of the mineralogy and phase chemistry of modern and ancient Gale crater dune fields, together with other measurements by Curiosity's science payload, provides new insights into present and past eolian processes on Mars.

Mathematical relationships between unit-cell parameters and chemical composition were developed for selected mineral phases observed with the CheMin X-ray diffractometer onboard the Curiosity rover in Gale crater. This study presents algorithms for estimating the chemical composition of phases based solely on X-ray diffraction data. The mineral systems include plagioclase, alkali feldspar, Mg-Fe-Ca C2/c clinopyroxene, Mg-Fe-Ca P2(1)/c clinopyroxene, Mg-Fe-Ca orthopyroxene, Mg-Fe olivine, magnetite, and other selected spinel oxides, and alunite-jarosite. These methods assume compositions of Na-Ca for plagioclase, K-Na for alkali feldspar, Mg-Fe-Ca for pyroxene, and Mg-Fe for olivine; however, some other minor elements may occur and their impact on measured unit-cell parameters is discussed. These crystal-chemical algorithms can be applied to material of any origin, whether that origin is Earth, Mars, an extraterrestrial body, or a laboratory.

Analyses by the CheMin X-ray diffraction instrument on Mars Science Laboratory show that gypsum, bassanite, and anhydrite are common minerals at Gale crater. Warm conditions (similar to 6 to 30 degrees C) within CheMin drive gypsum dehydration to bassanite; measured surface temperatures and modeled temperature depth profiles indicate that near-equatorial warm-season surface heating can also cause gypsum dehydration to bassanite. By accounting for instrumental dehydration effects we are able to quantify the in situ abundances of Ca-sulfate phases in sedimentary rocks and in eolian sands at Gale crater. All three Ca-sulfate minerals occur together in some sedimentary rocks and their abundances and associations vary stratigraphically. Several Ca-sulfate diagenetic events are indicated. Salinity-driven anhydrite precipitation at temperatures below similar to 50 degrees C may be supported by co-occurrence of more soluble salts. An alternative pathway to anhydrite via dehydration might be possible, but if so would likely be limited to warmer near-equatorial dark eolian sands that presently contain only anhydrite. The polyphase Ca-sulfate associations at Gale crater reflect limited opportunities for equilibration, and they presage mixed salt associations anticipated in higher strata that are more sulfate-rich and may mark local or global environmental change. Mineral transformations within CheMin also provide a better understanding of changes that might occur in samples returned from Mars.

We employ large mineralogical data resources to investigate the diversity and spatial distribution of vanadium minerals. Data for 219 approved species (http://http://rruff.info/ima, as of April 15, 2016), representing 5437 mineral species-locality pairs (http://http://mindat.org and other sources, as of April 15, 2016), facilitate statistical evaluation and network analysis of these vanadium minerals. V minerals form a sparse, moderately centralized and transitive network, and they cluster into at least seven groups, each of which indicates distinct paragenetic process. In addition, we construct the V mineral-locality bipartite network to reveal mineral diversity at each locality. It shows that only a few V minerals occur at more than three localities, while most minerals occur at one or two localities, conforming to a Large Number of Rare Events (LNRE) distribution. We apply the LNRE model to predict that at least 307 +/- 30 (1 sigma) vanadium minerals exist in Earth's crust today, indicating that at least 88 species have yet to be discovered-a minimum estimate because it assumes that new minerals will be found only using the same methods as in the past. Numerous additional vanadium minerals likely await discovery using micro-analytical methods. By applying LNRE models to subsets of V minerals, we speculate that most new vanadium minerals are to be discovered in sedimentary or hydrothermal non-U-V ore deposits other than igneous or metamorphic rocks/ore deposits.

The Letseng-la-Terae diamond mine in Lesotho, a small nation surrounded by South Africa, is unique in that it produces some of the world's largest and highest-value diamonds, from a relatively small volume of kimberlite ore. The mine, operated by De Beers during the late 1970s and early 1980s, was difficult to sustain economically because of its remoteness and very low ore grade, which made production costs very high. Since 2000, sharply rising prices for large diamonds have permitted the mine to reopen profitably as a 70/30 venture between Gem Diamonds Ltd. of South Africa and the government of Lesotho. To improve recovery of large diamonds, Letseng's owners have implemented new processing technology that provides better identification of these crystals before processing and a crushing mechanism that reduces their potential breakage. In recent years, several important diamonds recovered from Letseng have been sold to London luxury jeweler Laurence Graff.