The catastrophic atmospheric river (AR) events in California during December 2022-January 2023 highlight the need to understand high-risk weather extremes. This study uses machine learning to analyze AR clusters, finding that high-density clusters (high time fraction under AR conditions) exhibit more frequent high-category ARs, extreme precipitation, and severe land surface responses. Key circulation patterns are primarily attributed to subseasonal variability, modulated by daily geopotential height field variability. Climate model projections suggest more frequent high-density, high-category AR clusters with warming.

Yang Zhou, Michael Wehner, William Collins