Cold air outbreaks (CAOs) occur when large scale atmospheric circulations allow for the incursion of polar air masses into middle and lower latitudes, influencing wintertime temperatures regionally. The January 2004 CAO is identified as a major CAO in the Deep South of the United States in terms of wind chill equivalent temperature or a temperature-only criterion. Surface air temperature, horizontal winds, specific humidity, and Ertel potential vorticity are analyzed for this event using several reanalysis products: National Aeronautic and Space Administration Modern Era Retrospective-Analysis for Research and Application (MERRA), the National Centers for Environmental Prediction National Center for Atmospheric Research (NCEP-NCAR), and the National Centers for Environmental Prediction North American Regional Reanalysis (NARR). We perform an intercomparison of the reanalysis products and parallel surface station observations during the synoptic evolution of the leading cold front associated with CAO onset. The key synoptic, mesoscale, and dynamical features associated with onset are studied to determine the relative accuracy of the respective reanalysis products in representing the key features. The comparative evaluation revealed pronounced temperature and moisture biases in the NCEP-NCAR reanalysis products that limit its utility in portraying the synoptic features characteristic of CAO onset. Conversely, both MERRA and NARR accurately represent the detailed thermodynamic and moisture structural evolution associated with CAO onset indicating their utility in future observationally-based studies of CAO events. Ertel potential vorticity analyses indicate that the onset of the 2004 CAO is strongly linked to an incipient tropopause fold feature that developed over the Great Lakes region.

During the boreal cool season, regional climate in the United States is strongly impacted by extreme temperature regimes (ETRs), including both cold air outbreaks (CAOs) and warm waves (WWs), which have significant impacts on energy consumption, agriculture, as well as the human population. Using NCEP/NCAR and MERRA reanalysis data, the statistical characteristics of ETRs over three distinct geographical regions are studied: the Midwest (MW), Northeast Megalopolis (NE), and Deep South (SE). The regional long-term variability in the frequency and amplitude of ETRs is examined, and the modulation of these ETRs by low frequency modes is quantified. ETR behavior is characterized using three different metrics applied to both T and Twc: 1) the number of extreme cold/warm days, 2) a seasonal cumulative "impact factor", and 3) a peak normalized anomaly value. A trend analysis reveals a significant downward trend in SE WW events from 1949-2011. Otherwise, no significant trends are found for ETRs in any of the other regions. Thus, these results indicate that there has not been any significant reduction in either the amplitude or frequency of CAOs over the United States during the period of analysis. In fact, for the SE region, the recent winters of 2009/2010 and 2010/2011 both rank among the top 5 in terms of CAO metrics. In addition, strong interannual variability in ETRs is evident from 1949-2011 in each region. Linear regression analysis is then used to determine the associations between ETR metrics and the seasonal mean state of several low frequency modes, and it is found that ETRs tend to be modulated by certain low frequency modes. For instance, in the SE region, there is a significant association between ETRs and the phase of the North Atlantic (or Arctic) Oscillation (NAO/AO), the Pacific North American (PNA) pattern (for WWs only), the Pacific Decadal Oscillation (PDO) and the El Niño-Southern Oscillation (for WWs only). Over the MW region, WWs are modulated by the NAO/AO and PNA patterns, while in the NE region, the AO, NAO (for WWs only) and PDO (for WWs only) are implicated. In addition, it is found that there is an asymmetry between the low frequency mode modulation of CAOs and WWs. Multiple linear regression analysis is then used to quantify the relative roles of the various low frequency modes in explaining interannual variability in ETR metrics, and reveals that various combinations of low frequency modes can explain anywhere between 10% and 50% of the variance in the ETR metrics.