The next factor to consider is the three-dimensional nature of the atmosphere. We have seen how near-surface winds will tend to converge over the warmer SST (Fig. 1.4). As air comes together it must either descend or rise. Since it cannot go into the ocean, it rises. In other words, warmer SST tends to induce rising air. As air rises, so it cools and it may cool to the point where it can no longer hold moisture as a gaseous form. At this point, the moisture condenses into water droplets and clouds are formed. When water condenses from gaseous to liquid form, heat is released. We experience the reverse process when we step out of a swimming pool. Water on our bodies evaporates - from liquid to gaseous form - and this process absorbs heat - this makes us feel cold, until all water has evaporated and we are dry. The process of releasing heat when clouds are formed is important in the tropics because the increased warmth makes the air more buoyant and encourages further ascent. Around 10km, there is an effective lid on the atmosphere and air can no longer rise, even if it is releasing latent heat. At this point, air spreads out to either side and creates winds in a direction opposite to those at the surface (Fig. 1.4).

Fig. 1.4 describes a simple three-dimensional system of atmospheric circulation resulting from the direct effect of SST on the near-surface atmosphere. While it is a key process in generating seasonal to interannual climate anomalies, it is of course not the only factor at work. Indeed, the adage "what goes up must come down" is well founded in climate science. If we have air rising over warmer SST, then the air must descend somewhere. Descending air inhibits the development of clouds, so will tend to increase clear skies. This in turn can increase solar radiation during the day to heat up the SST. Thus, SST may rise as a result of descending air! In other words, it is possible for forcing from the upper levels of the atmosphere to influence the lower levels and the ocean. It turns out that this effect in typically not the dominant one, but illustrates how it is possible for the near-surface atmosphere in the tropics to depart from that expected from the Lindzen-Nigam model.