Many challenges exist with respect to the collection, storage, transfer, and analysis of "Big Data" across the Earth sciences. New Earth observing sensors and models generate richer and bigger datasets, and many methods are being used or developed to overcome the difficulty of working with them. This session discusses ongoing efforts to identify, assemble, and make large multivariate Earth datasets widely accessible for comparative AI research. Speakers will address real-world Big Data challenges they have encountered in their workflows and/or the solutions they have used to overcome them. Solutions could include data mining or machine learning algorithms, high-performance parallel computing systems or platforms, advanced data storage and transfer capabilities, and more.