The magnetic mineralogy of carbonaceous chondrites: a microscopy and machine learning study of meteorite WIS91600

Dataset 1: Full set of BSE images used to create Fig. 1 and for input to UNET segmentation. Each image tile is 250 µm x 250 µm wide. Images are organised into folders representing 43 columns of consecutive images (c_0 to c_42) that can be stitched to make the full image. Dataset 2: A collection of hyperspectral SEM/EDS datasets in .bcf format plus the Jupyter Notebook (Interactive_SEM_WIS.ipynb) to be run using SIGMA2 (see https://sigma2-documentation.readthedocs.io/en/latest/ to download and install SIGMA2). Files are named according to the corresponding Figured in the main text and supplement of the paper. Dataset 3: Full set of UNET segmented images derived from the BSE Dataset 1 used to create Fig. S1. Supplemental Dataset 4: Set of 10 selected BSE images (Dragonfly ORS file), their ground truth classification (Dragonfly ORS file), the modelled classification (Dragonfly ORS file) and the trained UNET model (.zip file for direct loading into Dragonfly). The data provided are those used to determine Table 1 and Fig. S4. Dataset 5: Files needed to recreate the micromagnetic simulation results presented in Figure. 13. Stich.pdf: A pdf showing the full stitched set of BSE images for FIg. 1. Yellow circles highlight many of the features focussed on in the paper. WIS Seg Stitch.pdf: A pdf showing the full stiched together segmented images. Zoom in to right lower corner to see evidence of impact texture. AGU Movie Test x: .avi and .mov versions of selected 3D animations of FIBnt results.