Multiple aspects of northern hemispheric wintertime cold extremes as revealed by Markov chain analysis

High-impact cold extremes have continued to bring devastating socioeconomic losses in recent years. In order to explain the exposure to cold extremes more comprehensively, this study investigates multiple aspects of boreal winter cold extremes, i.e., frequency, persistence, and entropy (Markovian descriptors). Cold extremes are defined by the bottom 10th percentile of daily minimum temperatures during 1950-2014 over the northern hemisphere. The spatial and temporal distributions of Markovian descriptors during 65 years are examined. Climatological mean fields show the spatial coincidence of higher frequency, shorter persistence, and higher entropy of cold extremes, and vice versa. In regard to the temporal variations over six representative regions of North America, Europe, and Asia, all regions share a decreasing tendency of frequency with the increases in regional winter mean temperature. By contrast, persistence and entropy show their intrinsic decadal variability depending on regions irrespective of the regional temperature variability, which give different information from frequency. Therefore, the exposure to cold extremes would not simply decrease with regional warming. Rather these results indicate that the descriptors with multiple aspects of the extremes would be needed to embrace the topical features as well as the holistic nature of cold extremes.