A Novel Geospatially Validated EMIT-Based Hyperspectral Dataset for AI-Driven Mineralogical Mapping

EMIT Hyperspectral Tabular Dataset for Supervised Mineral Classification over the multiple districts across Kabul Province and surrounding regions, including Kabul, Qala-e-Naim, Musayi, Khaki Jabar, Muhammad Agha, Tarakhel, Paghman, Mir Bacha Kot, Shakar Dara, Istalif, Mulla Muhammad Khel, Laghman, Nijrab, Alasay, and Jalrez. The Bagrami District chosen as the pilot region and independent spatial test area. This dataset supports supervised pixel-level mineral classification using hyperspectral reflectance data acquired by NASA's Earth Surface Mineral Dust Source Investigation (EMIT) imaging spectrometer, mounted aboard the International Space Station (ISS). The study area covers the Surobi district of Afghanistan, a geologically complex region with diverse surface mineralogy. The dataset is provided in tabular (CSV) format for direct compatibility with machine learning and Deep learning pipelines (e.g., scikit-learn, PyTorch, TensorFlow). Each row represents one pixel, with 244 spectral band reflectance values as featuresand a single integer mineral class label. Ground truth labels were derived from USGS geological reference maps.