A hyperspectral machine learning image library for ice core feature identification

Here, we share the first edition of an image library for training a machine learning model designed to identify air bubbles and other ice features in ice cores. This datset consists of annotated ice cores images from Allan Hills (ALHIC2302), South Pole Ice Core (SPICE) project (SPC-14), Greenland Ice Sheet Project 2 (GISP2D), West Antarctic Ice Sheet (WAIS) Divide (WDC-06A), and Denali (DEN13). The intention behind the dataset is to lay the foundtaions for an image library that leverages marchine learning to enhance ice feature detection, allowing for studies of bubble size and orientation and/or inter-bubble space.