Mineralogical Classification of CRISM Hyperspectral Data Under Uncertainty With Hybrid Neural Networks

Orbital remote sensing observations are a lynchpin of planetary science research. Hyperspectralinfrared spectroscopy in particular is key for planetary mineralogical exploration, for example, CRISM forMars, as this underpins our understanding of the distribution of specific lithologies and the geological processleading to their formation. Yet routine analysis workflows involving summary parameters have significantlimitations and are highly time‐consuming. This work presents a novel methodology and framework for theanalysis and classification of CRISM SWIR reflectance spectroscopy, leveraging Machine Learning (ML). Wetrain a model to classify 37 minerals previously manually identified on the planet. We show this model is highlyperformant, with test data across Mars and a case study within Jezero crater, where ML results match previousmanual analyses and rover observations. We also adapt Expected Cost (EC) to remote sensing data for use ingeological context for the first time. We demonstrate that EC can be used to dynamically weightmisclassification penalties based on geological context, as a rigorous measure of automated classificationmethods. We envision this model to make analysis of CRISM data more accessible to the planetary sciencecommunity, allowing rapid searches for a vast range of minerals across a global/regional scale. As a result, areasof interest for further satellite or rover exploration can be quickly identified, leading to greater understanding ofgeological processes on Mars