Predictors and prediction skill for marine cold‐air outbreaks over the Barents Sea

• Published in: Quarterly journal of the Royal Meteorological Society Vol. 147; no. 738; pp. 2638 - 2656
• Main Authors: Polkova, Iuliia, Afargan‐Gerstman, Hilla, Domeisen, Daniela I. V., King, Martin P., Ruggieri, Paolo, Athanasiadis, Panos, Dobrynin, Mikhail, Aarnes, Øivin, Kretschmer, Marlene, Baehr, Johanna
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.07.2021; Wiley Subscription Services, Inc
• Subjects: Arctic climate; Atmospheric circulation; Atmospheric circulation patterns; causal drivers; marine cold air outbreak; Outbreaks; polar low; Polar lows; Predictions; Sea surface; Sea surface temperature; seasonal prediction; Surface temperature
• DOI: 10.1002/qj.4038
• ISSN: 0035-9009

Marine cold‐air outbreaks (MCAOs) create conditions for hazardous maritime mesocyclones (polar lows) posing risks to marine infrastructure. For marine management, skilful predictions of MCAOs would be highly beneficial. For this reason, we investigate (a) the ability of a seasonal prediction system to predict MCAOs and (b) the possibilities to improve predictions through large‐scale causal drivers. Our results show that the seasonal ensemble predictions have high prediction skill for MCAOs over the Nordic Seas for about 20 days starting from November initial conditions. To study causal drivers of MCAOs, we utilize a causal effect network approach applied to the atmospheric reanalysis ERA‐Interim and identify local sea surface temperature and atmospheric circulation patterns over Scandinavia as valuable predictors. Prediction skill for MCAOs is further improved up to 40 days by including MCAO predictors in the analysis. Marine cold‐air outbreaks (MCAOs) create conditions for hazardous maritime mesocyclones known as polar lows posing risks to marine infrastructure. For marine management, prediction of MCAOs would be highly beneficial. The seasonal prediction system is able to predict MCAOs up to 20 days. By combining statistical and dynamical predictions, this skill can further be extended to 40 days in some regions.