Abstract: Clouds play a critical role in the Earth’s energy budget and their potential changes are the largest source of uncertainty in future climate projections. However, the use of satellite observations to understand cloud feedbacks in a warming climate has been hampered by the simplicity of existing cloud classification schemes, which are based on single-pixel cloud properties rather than utilizing spatial structures and textures. Recent advances in computer vision enable the grouping of different patterns of images without using human-predefined labels, providing a novel means of automated cloud classification. This unsupervised learning approach allows discovery of previously unknown climate-relevant cloud patterns, and the automated processing of large datasets.

I describe here the new AI-driven Cloud Classification Atlas (AICCA), which leverages 22 years and 800 terabytes of MODIS satellite observations over the global ocean. We use a rotation-invariant cloud clustering (RICC) method to classify those observations into 42 AI-generated cloud class labels at ~100 km spatial resolution for approximately 160M observed cloud classes. I also highlight new climate science findings from AICCA, namely the meteorological controls and dynamics for low-altitude clouds in the subtropics with some implications for future climate change.

Co-authors: Takyua Kurihana, Ziwei Wang, Ian Foster, Liz Moyer

Bio: Jim Franke is a 6^th year PhD student in Geophysical Sciences at the University of Chicago. His research focuses on the impacts of climate change on human and natural systems at the intersection of climate and computer sciences. Prior to UChicago, Jim received a bachelor’s degree in Mechanical Engineering from the Milwaukee School of Engineering and was a consulting energy engineer at Environmental Systems Design in Chicago.