Using the self-organizing map algorithm to characterize widespread temperature extreme events over Alaska

The spatial resolution of global climate models is such that localized extreme weather events are difficult to simulate. Thus, relating the large-scale synoptic conditions (a feature that global climate models reproduce well) to localized extreme events may be useful for predicting the frequency, timing, and locations of such events. This study focuses on widespread warm and cold temperature extremes (defined as the warmest and coldest 1% of all temperatures occurring on at least 25 grid points on a given day) within two domains in Alaska divided by the Brooks Range. The synoptic climatology for Alaska is defined using daily sea level pressure (SLP) data from the ERA-Interim reanalysis, which is categorized into 35 patterns using the self-organizing map (SOM) algorithm. The SOM classification is used to relate these large-scale SLP circulation patterns to local extreme temperature events. Additionally, the preceding five days leading up to extreme events are examined through the SOM classification. Extreme events may be related to synoptic blocking patterns, leading to high or low temperatures caused by advection or radiative heating or cooling. Observing the days preceding extreme temperatures may shed light on the influence of such blocking patterns.