Utilizing a Self-Organizing Map to Identify Synoptic Patterns in Heavy Precipitation Events in the Northeastern United States

Self-Organizing Maps (SOMs) have been used extensively across multiple fields to classify or cluster information contained in large datasets. However, until recently, SOMs were not widely used in the field of climatology. Climate researchers are increasingly transitioning from more manually intensive identification processes to more automated machine learning approaches. This study utilizes the ERA5 climate reanalysis dataset from 1979–2018 to identify heavy precipitation days in the Northeastern US. A SOM is then applied to the sea-level pressure patterns of these days to classify archetypal synoptic patterns associated with heavy precipitation events. This classification can further enhance our understanding of the causes of heavy precipitation events in the context of their synoptic patterns.