A methodology for tracking cold spells in space and time: development, evaluation and applications

Cold spells, identified as periods of prolonged extreme low temperatures, are often analysed in an Eulerian framework or through the use of case studies. However, this restricts information about their spatio-temporal evolution and limits the ability to compare analogous events that share similar developments. This study identifies cold spells as a series of geographical objects that are connected across subsequent time-steps. These objects are characterised by persistent low-temperature anomalies and can be grouped into the same event by using a connected component method, previously applied to heatwaves. This work extends this method further by taking into account advection by the tropospheric mean (weighted vertical average) horizontal wind. We also extend the methodology to filter quasi-stationary events that may have different drivers to transient events. Once catalogued, the cold spells are easily accessed based on their properties, location, or time period for further study. This study applies the cold spell identification methodology to the European Centre for ECMWF Reanalysis v5 (ERA5) dataset to develop a climatology of cold spells in the Northern hemisphere, establishing their seasonal variations and associated patterns of atmospheric circulation. We compare the results to an existing Eulerian based methodology and explain some reasons for differences in the cold spells identified. This study also demonstrates the typical pathways by which cold spells reach regions of East Asia, Europe and North America.