Data for: "X-PRISM: An Open-Source Spatial-Spectral Workflow for Automated Mineral Mapping from Micro-XRF"

This dataset supports the case study and quantitative model evaluation presented in “X-PRISM: An Open-Source Spatial-Spectral Workflow for Automated Mineral Mapping from Micro-XRF Data.” It is distributed as two archives. xprism_case_study_data.zip This archive contains four raw Bruker BCF hyperspectral mapping files: 1-dl.bcf, 1-dr.bcf, 1-ul.bcf, and 1-ur.bcf. Each BCF file contains spatially registered micro-XRF spectra acquired using a Bruker M6 JETSTREAM spectrometer. Data-acquisition parameters were: Spot size: 50 μm Step size: 160 μm Dwell time: 15 ms per pixel Tube voltage: 50 kV Tube current: 600 μA These files provide the raw hyperspectral micro-XRF measurements used in the Cu–Mo skarn case study. xprism_independent_benchmark_data.zip This archive contains a compressed NumPy (.npz) file used for quantitative model evaluation. The file contains two principal fields: hypermap: micro-XRF hyperspectral data, with a full spectrum associated with each spatial position. label: spatially corresponding integer reference labels used for performance evaluation. Class encoding: −1: Background 0: Calcite 1: Chalcopyrite 2: Andradite 3: Molybdenite 4: Pyrite 5: Quartz The benchmark data can be loaded in Python using NumPy: data = numpy.load("FILE_NAME.npz")hypermap = data["hypermap"]labels = data["label"] Background pixels encoded as −1 should normally be excluded when calculating performance metrics.