In this research, we define extreme temperature events using a recently defined excess heat factor, based on the exceedance of apparent temperature beyond the 95th percentile along with an acclimatization factor, to define extreme heat events (EHE). We extend the calculation to assess cold and develop relative metrics to complement the absolute metrics, where extremeness is based on conditions relative to season. We thus examine extreme cold events (ECE), relative extreme heat events, and relative extreme cold events in addition to EHE. We present a climatology of these variables for North America, followed by analyses of trends from 1980 to 2016. While EHE and ECE are found in the core of summer and winter, respectively, relative events tend to have a broader seasonality. Trends in relative extreme heat events and EHE are upward, and relative extreme cold events and ECE are downward; the relative events are changing more rapidly than the absolute events.Plain Language Summary One of the most critical ways in which weather conditions influence the environment is through extreme temperature events. While excessive heat and cold conditions have been amply studied, events that are extreme relative to the time of year have been less examined. These relative events may grab fewer headlines but can have important impacts on the environment, agriculture, and human health. In this research, we present a climatology of cold and heat events, both absolute and relative, for North America, followed by an analysis how they have changed from 1980 to 2016. Results show an increase in heat events and decrease in cold events across most of the United States and Canada. More interestingly, the relative events are changing slightly more rapidly than the absolute events.
SHERIDAN AND LEE11,891 subsequent k-means cluster analysis. This process resulted in each grid point being categorized into one of 12 regions. Multiple numbers of regions (between k = 5 and k = 15) were examined, and ultimately 12 were chosen based upon qualitative assessment as well as 12 regions having the best variability skill score (Lee, 2014). These statistically derived regions were then used to inform a customized drawing of the final regions in ArcMap.