THRawS is a new dataset of raw Sentinel-2 (S-2) satellite data containing warm temperature hotspots such as wildfires and volcanic eruptions from around the world. The dataset aims to promote the development of energy-efficient pre-processing algorithms and AI models for onboard-satellite applications. A custom methodology was designed to identify events in raw data using corresponding Level-1C (L1C) products and a lightweight coarse coregistration and georeferencing strategy was employed to deal with unprocessed data. The dataset comprises over 100 samples, including wildfire, volcanic eruption, and event-free volcanic areas, to enable warm-events detection and general classification applications. Finally, the performances of the proposed coarse spatial coregistration technique and the SuperGlue Deep Neural Network method were compared to highlight different constraints in terms of timing and quality of spatial registration to minimize spatial displacement error for a specific scene. Github Repo: https://github.com/ESA-PhiLab/PyRawS

- The json file includes the corresponding L1C files. - Github Repo:https://github.com/ESA-PhiLab/PyRawS