In summary, anomalies of SST in the tropics can induce large atmospheric circulation and heating anomalies through the depth of the local atmosphere. By modifying the tropical climate patterns over the tropical oceans, the atmospheric patterns are often quite strongly affected over tropical land masses. In addition, the changes in the tropical atmosphere can set off wave chains into higher latitudes, so some predictability is found at these latitudes too, especially close to the Pacific basin where the forcing on the climate system from SST variations is strongest, most clearly in association with El Niño.

The ocean-atmosphere climate system functions quite differently away from the tropics - mainly because the Coriolis effect is much stronger. The reason for this can be seen if we view the earth side on. In the deep tropics, moving a distance of 500km polewards takes us into a region where the circumference of the earth is changed only marginally - that is, it will be moving at only a slightly slower speed relative to where we started. However, in mid-latitudes, moving 500km toward the pole takes into a region with substantially smaller circumference and thus, will be moving substantially slower than the point where we started, just 500km away. The net result is that the Coriolis force becomes a more dominant player in setting up the balance of forces and resulting winds in Mid-latitudes, and large gradients of ocean surface temperature generate much smaller winds than in the tropics. In contrast, mechanisms internal to the atmosphere often related to the high Coriolis force, are able to generate strong seasonal climate anomalies. For example, the North Atlantic Oscillation describes large swings in the strength of westerly winds across the North Atlantic and the character of storms that enter into Europe and NW Africa, especially during winter time. Very little of the year-to-year variation in the North Atlantic Oscillation appears attributable to SST forcing.

The above discussion has focused on year-to-year variations of climate. It may be useful to bring to mind at this point that climate also varies on multi-year timescales as well, for which SST forcing in the tropics may also play a role, but along with many other processes. In particular it is now clear that climate naturally varies on decadal timescales. The causes and predictability of climate fluctuations on these timescales is being actively investigated. Another much researched aspect is the multidecadal to century timescale response of the climate system to forcings initiated by human activity.

There are now many examples of detailed studies of the relationship between regional climate over land areas and the ocean-atmosphere variations - under the hypothesis that the variations of SST are partly forcing the regional climate variations over land, based on the concepts developed in the first part of this lecture. We will often use the example of East Africa's October-December rainfall season to illustrate concepts. This region's rainfall has a strong association with ocean-atmosphere variability, as indicated in the schematic in Fig. 1.8. To illustrate the types of analysis that are made, Fig 1.10 shows the area average rainfall observed in each October-December season in the region, while Fig. 1.11 shows the mean SST, near-surface wind, and rainfall anomaly patterns that were observed during the five wettest seasons in East Africa. The features shown in the schematic in Fig. 1.8 show up well in the actual observed fields during the above normal rainfall years. In addition to the Pacific and Indian Oceans, the map also suggests that there may be a role for forcing from the tropical Atlantic (warmer than normal SST in the Equatorial and southern tropical Atlantic during the wet years). However, it is difficult to prove these connections based on statistical relationships alone, especially when the relationships are weak, as there is always the risk of finding relationships that are merely the result of statistical coincidence, rather than physical connections in the climate system. Therefore, these diagnostic studies of observations can be reinforced by studies of climate variability in comprehensive climate models (General Circulation Models - GCMs), as discussed below.