Cool Season Temperature Regimes: Low Frequency Modulation and Representation in CMIP5 Models

Abstract
During the cool season regional climate is strongly impacted by anomalous temperature regimes (TRs) including both cold air outbreaks (CAOs) and warm waves (WWs), which have significant impacts on energy consumption, agriculture, and the human population. Using reanalyses and CMIP5 output, the statistical behavior of TRs impacts are studied concentrating on three geographical regions in the United States: the Midwest (MW), Northeast (NE), and Southeast (SE). TR behavior is characterized using a new cumulative seasonal “impact factor” metric. The regional long-term TR variability is examined and the modulation of ETRs by low frequency modes is quantified. The observed behavior is then contrasted with the parallel analyses of historical CMIP5 simulations.
Trend analyses reveal only a marginally significant decrease in SE WW events from 1949-2011. Otherwise, no significant trends are identified for TRs in any of the regions considered. Thus, there is no significant reduction in the impact of CAOs over the United States over recent decades. Nonetheless, robust interannual variability in TRs is evident within each region and statistical analysis reveals that this behavior is due, in part, by the seasonal modulation of TRs by several prominent natural modes of low frequency variability including the North Atlantic Oscillation (NAO or AO), Pacific-North American, Pacific Decadal Oscillation and the El Nino-Southern Oscillation. Although asymmetries exist between the low frequency modulation of CAOs and WWs, multiple linear regression analyses indicate that the collective influence of the low frequency modes accounts for as much as 50% of the regional interannual variability in TR impact. Model validation research is composed of assessing (a) regional TR behavior, (b) the representation of low frequency modes, themselves and (c) TR-low frequency mode linkages. A key result is that serious deficiencies are uncovered in the representation of major low frequency modes by CMIP5 models.