A modelling study of the impact of tropical SSTs on the variability and predictable components of seasonal atmospheric circulation in the North Atlantic–European region

Atmospheric variability and predictable components over the North Atlantic–European region (NAE) were analysed in the late-winter season using a general circulation model of intermediate complexity (ICTP AGCM). The method of empirical orthogonal functions (EOF) analysis was applied to 200-hPa geopotential heights to extract individual modes of variability occurring in the ensemble of numerical simulations. The same variable was selected for the signal-to-noise optimal patterns method, which identifies the patterns maximising the signal-to-noise ratio, following Straus et al. (J Clim 16(22):3629–3649, 2003). Six experiments based on a 35-member ensemble of 156-year long simulations were conducted to detect the potential impact of tropical sea surface temperatures. Each experiment was forced with observed sea surface temperature anomalies prescribed in different ocean areas: globally, in the entire tropical zone, the tropical Atlantic region, the tropical Pacific area, and the tropical Indian Ocean area. In late winter, the leading EOF pattern calculated for all individual ensemble members projects onto the North Atlantic Oscillation, while the second EOF pattern projects onto the East Atlantic pattern. However, EOF modes based on the ensemble mean, which should reflect the forced component of the signal, have different spatial characteristics. Alongside the classical analysis of signal and noise, results of the signal-to-noise optimal patterns method suggest that the optimal patterns and signal-to-noise ratio are affected by the boundary forcing of the oceans. Furthermore, the resemblance between the first optimal pattern and the EOF1 pattern based on the ensemble mean points toward the vital role of the lower-boundary-forced signal in establishing potential predictability in the NAE region.