Iceland's Great Frost Winter of 1917/1918 and its representation in reanalyses of the twentieth century

• Published in: Quarterly Journal of the Royal Meteorological Society
• Main Authors: Moore, G. W. K., Babij, M.
• Other Authors: Natural Sciences and Engineering Research Council of Canada
• Format: Article in Journal/Newspaper
• Language: English
• Published: Wiley 2016
• Subjects: Arctic; greenlandic; Iceland; North Atlantic; North Atlantic oscillation; Sea ice
• Online Access: http://dx.doi.org/10.1002/qj.2939; https://api.wiley.com/onlinelibrary/tdm/v1/articles/10.1002%2Fqj.2939; https://rmets.onlinelibrary.wiley.com/doi/pdf/10.1002/qj.2939

Climate variability during the twentieth century is a subject of considerable interest as it represents a way to place the changes that we are currently experiencing into a long‐term context. The characterization of climatic extremes and their representation in models is of particular importance as these events have a significant impact on communities and ecosystems. Much of the focus on these extremes has been on events that occurred after the establishment of the upper‐air observing network in the 1950s. The recent availability of reanalyses that extend throughout the twentieth century now allow extreme events that occurred before this time to be more fully investigated. The winter of 1917/1918 is referred to as the Great Frost Winter in Iceland. It was the coldest winter in the region during the twentieth century. It was remarkable for the presence of sea ice in Reykjavik Harbour as well as for the unusually large number of polar bear sightings in northern Iceland. Here we use observations as well as two reanalyses that span the twentieth century to document this event. We show that throughout much of the region, January 1918 was the coldest winter month on record. The North Atlantic Oscillation (NAO) attained one of its most negative values during January 1918 and the westward shift in its northern centre of action allowed cold Arctic and Greenlandic air to penetrate south towards Iceland. We also show that the two reanalyses diverged in their ability to represent the temperature anomalies during the event. Differences in the sea‐surface temperature (SST) and sea ice concentration fields used to force the underlying models contributed to this divergence. These results stress the important role that boundary conditions play in determining the fidelity with which reanalyses can represent extreme climate events in data‐sparse regions.