Advancements in ARM User Facility Products and Tools using Machine Learning to Support Atmospheric Research

New observational datasets of atmospheric state and key atmospheric processes and quantifying observational uncertainties are essential to better understand different feedback mechanisms and increase the fidelity of models at different scales. The U.S. Department of Energy Atmospheric Radiation Measurement (ARM) User Facility aims to alleviate these and other deficiencies and needs by providing tools and comprehensive suites of advanced in-situ and remote-sensing ground-based and airborne observations. Here, we present new and updated retrievals and high-level data products developed at ARM, leveraging machine learning (ML) and other advanced techniques. These ML-augmented multi-instrument retrievals provide useful microphysical quantities, accompanied by uncertainty estimates, including ice precipitation microphysical properties in sub-cloud profiles, hydrometeor phase classification profiles, all-sky imager pixel segmentation, and aerosol size distributions spanning an extensive size spectrum. Finally, we also present a set of ARM-supported tools to bridge between ARM observations and model simulations, such as the Earth Model Column Collaboratory (EMC²).