Assessment of the quality of ensemble seasonal forecasts in the Arctic region
Keywords:
Arctic, Arctic oscillation, ensemble forecast, quality assessment, air temperature, mean sea level pressure, CFSv2.Abstract
The Arctic region plays a crucial role in the Earth’s climate system, and the demand for reliable seasonal forecasts in this area is increasing under ongoing climate change. The aim of this study is to assess the quality of ensemble seasonal forecasts of key meteorological variables in the Arctic and to identify the factors controlling their predictability.
The analysis is based on retrospective ensemble forecasts produced by the CFSv2 climate forecasting system with lead times of up to three months. Forecasts were verified against atmospheric reanalysis data. Forecast skill was quantified using two complementary metrics: the correlation coefficient between predicted and observed anomalies and a parameter characterizing the accuracy of reproducing the sign and spatial structure of these anomalies.
The results show that forecasts of mean sea level pressure demonstrate higher and more stable predictability than forecasts of near-surface air temperature, particularly at lead times of two to three months. Forecast skill exhibits pronounced seasonality, reaching its maximum during the cold season and decreasing markedly in summer. It is also shown that positive forecast skill at a one-month lead time can serve as an indicator of the persistence of predictive information at longer lead times.
A clear regime dependence is identified: forecast skill, especially for the pressure field, is statistically higher during the positive phase of the Arctic Oscillation. These results provide a basis for practical recommendations on the interpretation and adaptive use of seasonal forecasts in the Arctic, taking into account seasonal predictability, circulation regimes, and the relative informativeness of different meteorological variables.
