Impact of tropical teleconnections on the long-range predictability of the atmosphere at midlatitudes: a reduced-order multi-scale model perspective

Abstract

Abstract Teleconnections between the tropical and the extratropical climates are often considered as a potential source of long-term predictability at seasonal to decadal time scales in the extratropics. This claim is taken up in the present work by investigating the predictability of a coupled ocean–atmosphere extratropical model under a one-way forcing generated by a tropical model. Both models display a chaotic dynamics, and the dominant variable of the extratropical model displays a high correlation with the tropical forcing in the reference simulation, inducing a low-frequency variability signal in the extratropics. Numerical experiments emulating the presence of initial condition errors in the tropical model are conducted to clarify their impact on the predictability in the extratropics. It is shown that: (i) the correlation skill of the dominant observable affected by the forcing is considerably degraded at interannual time scales due to the presence of initial condition errors in the tropics, limiting the potential of teleconnections; (ii) averaging of an ensemble of forecasts–with a small number of members–may substantially improve the quality of the forecasts; and (iii) temporal averaging may also improve the quality of the forecasts (at the expense of being able to forecast extreme events), in particular when the forcing affects weakly the observable under interest.