Prediction of Northern Hemisphere Regional Surface Temperatures and the Cryosphere using Stratospheric Ozone Information

Correlations between springtime Arctic stratospheric ozone extremes and subsequent surface temperatures have been previously reported for both models and observations at particular locations in the Northern Hemisphere. Motivated by the routine and precise measurements of total column ozone, as well as ozone’s ability to act as a proxy for the wintertime polar vortex, here, we quantify the use of ozone information for seasonal forecasts of Northern Hemisphere temperatures and the cryosphere using both observations and a nine-member ensemble of the The Community Earth System Model, version 1 Whole Atmosphere Community Climate Model, version 4. The observed and ensemble composite correlations between March Arctic total column ozone and April surface temperatures show remarkable consistency throughout the Northern Hemisphere. Additionally, large surface temperature differences, up to 6 degrees in Northern Russia, are also seen between composite years of opposing ozone extremes. Similar connections between March total column ozone and Northern Hemisphere sea ice and snow cover are observed. However, remarkably, sea ice extent differences show persistence through late autumn and early winter, especially in the Laptev, East Siberian, Chukchi, and Beaufort Seas, and in the Bering strait. Snow depth is largely impacted in the Northern Russian region where surface temperature differences were greatest. However, the largest differences are occurring along the South East Coast of Alaska in early summer before the snow cover diminishes. To investigate these connections in more detail, we create an empirical forecast model using March Arctic total column ozone to predict the sign of the observed, as well as the modeled surface temperatures, sea ice and snow depth anomalies from April-July for temperatures and April-December for the cryosphere. This is done through a leave-three-years-out cross validation method, which produces, for example, up to 75% correct predictions of April surface temperature anomalies in the model ensemble in Northern Russia.