Although infrequent, ice storms represent significant high-impact winter weather events that can lead to severe disruptions in transportation and energy infrastructure, resulting in extensive socio-economic ramifications. Assessing the capability of climate models to accurately simulate frozen precipitation extremes is critical for enhancing model fidelity and increasing stakeholder and policymaker confidence in future climate projections. However, a predominant limitation of most climate models is their inability to differentiate precipitation types beyond snow and rain, which poses a challenge for evaluating the impacts associated with ice accumulation and future changes.

In this presentation, we leverage three-dimensional thermodynamical fields produced by global models to implicitly define frozen precipitation classes (sleet, freezing rain, and snow) by implementing an ensemble precipitation typing approach using the Bourgouin, Ramer, and Revised techniques, typically applied to numerical weather prediction output. We validate this approach with ERA5 and station observations from the Automated Surface Observing System (ASOS) and NOAA’s Integrated Surface Database (ISD) across the northeastern U.S. (NEUS) during 1980-2019. We further employ self-organizing maps (SOMs) as an automated machine-learning approach to characterize the large-scale meteorological patterns (LSMPs) associated with freezing rain events in the region. Trained on ERA5 dynamic and thermodynamic fields, we define atmospheric setups critical for a model to capture in order to accurately simulate frozen extremes. The reanalysis-derived SOM is then applied as a reference in order to assess the skill of CMIP6 historical experiments in simulating the synoptic environments conducive to freezing rain events in the mid-Atlantic. We discuss how these methods could illustrate how model biases arise from a model's thermodynamic or dynamic behavior and how such tools could provide insight into future changes in freezing rain risk over eastern North America.