Environmental and Mineralogical Characteristics of Tourmaline-bearing localities in the United States

Background This dataset contains the compiled mineralogical records of 458 tourmaline-bearing sites in the United States. Environmental parameters associated with each locality such as surface temperature, precipitation, and geothermal gradient have also been recorded. Geological processes were identified for each locality using site and mineral descriptions from mineral databases and they were also interpreted using machine learning algorithms. Market basket analysis algorithms were utilized to identify patterns within the mineralogical data and these patterns were attributed to the geological processes listed in the dataset. Data Sources Mineral occurrence data were retrieved from Mindat.org for the tourmaline minerals schorl, dravite, elbaite, uvite, and foitite. Mineral descriptions listed in the Handbook of Mineralogy (Anthony et al., 2004) web database aided in the identification of different geological processes. Geothermal gradient data are interpolated from values listed in geothermal temperature studies conducted by Batir et al. (2013), Kron and Stix (1982), and Nathenson and Guffanti (1987). Elevation data were retrieved from the USGS Elevation Point Query Service and the climatic data from the WorldClim dataset. Köppen-Geiger climatic zones were assigned to each locality using data published by Kottek et al. (2006) and made available at http://koeppen-geiger.vu-wien.ac.at/ Version Two The second version of this dataset (TourmalineLocalitiesV2.csv) contains the initial tourmaline-bearing sites retrieved from the Mindat mineral database plus 389 randomly generated test localities. The randomly generated locations have all of the same parameters as the initial dataset with the mineralogy being extracted from mineral surface models. Natural neighbor interpolation was used to generate surface models for each individual mineral in the dataset and the mineralogy for each test locality was extracted from the models. The first version of the dataset also did not have coordinates for each site and those have been added in this version.