Are springtime Arctic ozone concentrations predictable from wintertime observations?

The northern hemisphere stratospheric polar vortex, and thus Arctic column ozone content, is characterized by large interannual variability, driven by the interplay of various chemical, dynamical and meteorological forcings throughout the winter and spring season. The 2023/24 season showed record high March total column ozone, whereas 2010/11 and 2019/20 experienced large springtime Arctic ozone losses due to an exceptionally strong and prolonged polar vortex state. Another remarkable season occurred during 2015/16, where unprecedented cold stratospheric temperatures in January were interrupted by a sudden stratospheric warming event and thus the fear of large springtime ozone losses did not manifest. Our main research question is motivated by these events: To which extent can springtime Arctic ozone concentrations be predicted from the preceding wintertime observational record? To this end we investigate the suitability of wintertime mean polar cap temperature, PSC proxies and eddy heat flux as predictors of springtime ozone in ERA5 and MERRA2 reanalysis data. Our results show that using these predictors springtime ozone can only be ‘forecast’ with short lead times, and even then with limited accuracy. In contrary expanding the analysis to ozone observations earlier in the season, we find substantially higher predictive skill compared to temperature, PSC proxies or eddy heat flux, which can be understood as ozone reflecting both the chemical and dynamical conditions over the northern polar cap.