North American cold air outbreaks (CAOs) can have devastating socio-economic impacts. Typically these CAOs are identified using Eularian criteria in fixed regions or studied using specific case studies. In order to better represent the changing shape, size, and location of CAOs across the continent, we employed TempestExtremesV2.1 (TE), a Lagrangian feature tracking method. TE enabled us to identify and track all CAOs in the European Center for Medium-Range Weather Forecasts reanalysis dataset (ERA5) for DJF 1979-80 to 2020-21. CAOs were detected based on detrended, standardized anomalies relative to the daily climatology in temperature at 2-meters (T2m). CAOs were also identified in the Community Earth System Model (CESM) Large Ensemble 2 (LENS2) and the ERA5 reanalysis re-gridded to the LENS horizontal resolution. Additionally, a quasi-Lagrangian thermodynamic energy budget was used to identify the role of horizontal temperature advection, adiabatic processes, and diabatic processes in driving CAO development in both datasets. Events were then generalized based on the dominant mechanisms contributing to warming and cooling within the CAO as well as the lysis location. This enabled a process-oriented model evaluation of CAO events. A comparison of characteristics such as duration, intensity, and number of cases detected at a given location was conducted between the ERA5 and LENS2 CAOs. Furthermore, we conduct an investigation into potential biases in processes occurring within these events. In addition, the availability of numerous ensemble members in the LENS2 enables a robust analysis of trends in CAOs over the historical period, taking into account internal variability.