We consider the mechanisms through which the ocean provides signals which are sensitive for atmospheric variability on different time scales. A critical question for climate modelling and climate prediction is how to discriminate anthropogenic signals from natural variability. Different climate modelling and diagnostics groups discovered that on multidecadal time scales, the ocean drives predictable climate variability that superimposes the long-term anthropogenic trend. However, to further explore the mechanisms we need to analyse the long-term variability in air-sea energy fluxes - the language by which ocean and atmosphere communicate. For this purpose we developed innovative statistical tools to compute the air-sea fluxes using limited amount of data, This enabled us to prove the Bjerknes' conjecture that the ocean drives climate variability in the North Atlantic sector on time scales of decadal and longer, formulated nearly 50 years ago. The results provided a new dimension for understanding the role of natural variability associated with the ocean during the 20th century and may lead to more accurate climate prediction. In the next instance we analyzed the mechanisms responsible for forming surface turbulent fluxes and their potential role in changes of midlatitude heat balance. Midlatitude cyclones were considered as the major mechanism responsible for extreme turbulent fluxes which are typically occur during the cold air outbreaks in the rear parts of cyclones when atmospheric conditions provide locally high winds and air-sea temperature gradients. For this purpose we linked characteristics of cyclone activity over the midlatitudinal oceans with the extreme surface turbulent heat fluxes. Cyclone tracks and parameters of cyclone life cycle (deepening rates, propagation velocities, life time and clustering) were derived from the same reanalyses using state of the art numerical tracking algorithm. Our analysis has shown that extreme surface fluxes are typically associated not with cyclones themselves but rather with cyclone-anticyclone interaction zones. This implies that North Atlantic and North Pacific series of intense cyclones do not result in the anomalous surface fluxes. Alternatively, extreme fluxes are most frequently associated with blocking situations, particularly with the intensification of the Siberian and North American Anticyclones providing cold-air outbreaks over WBC regions.