At this point it is useful to summarise the previous sections in terms of a single coherent model of the world's circulation overall. The model must account for the distribution of pressures (Figure 1.8), the meridional transfer of heat (Note 5.F) and the latitudinal variations of rainfall (Figure 10.6) and winds (Figure 12.5).
Various models have been suggested in the past. Edmund Halley proposed in 1686 that the easterly Trade winds were following the Sun, flowing towards the part of the Earth that is warmed by solar radiation, like the draught to a fire. But that would imply a daily reversal of winds.
A better model was advocated by George Hadley in 1735. He explained the Trades as due to a lag of the wind on the rotation of the Earth: the ground moves eastwards faster than the atmosphere does, as the air is drawn towards the equator by the warmth there. Then there is thermal convection upwards from the equator, followed by subsidence at higher latitudes. This explains the pattern within the Hadley cells (Section 12.3) and contains an early interpretation of the Coriolis effect, but fails to account for surface westerlies at higher latitudes.
That objection was later considered by William Ferrel in the light of the Coriolis effect (Note 11.D). He suggested that the Hadley cell meshes with a midlatitude cell rotating in the reverse direction, which in turn interlocks with a polar cell beyond, like three cog-wheels in a row. Such a three-cell model of the general circulation was more clearly described by Tor Bergeron in 1928, and was used in Section 12.3 in considering Figure 12.10 and Figure 12.11. It has the advantage of being simple, but has at least three difficulties. It implies a boundary between the Ferrel and polar cells at latitudes much higher than is observed in practice; polar air actually meets low-latitude warm air at fronts nearer the equator than 60°. Secondly, the three-cell model implies upper easterlies above the midlatitudes (as the counterpart of the surface westerlies), which is incorrect (Figure 12.10). Thirdly, the circulation known as the Ferrel cell proves to be insignificant. In fact, it is now realised that circulations in midlatitudes are not regular flows around horizontal axes (as in the three-cell model), but are largely the outcome of sporadic, asymmetric eddies around vertical axes.
So we arrive at a more complex version (Figure 12.16), following the ideas of Rossby (1941), Palmen (1951) and Newton (1969), and incorporating what has been learnt about the upper winds. This model almost omits the Ferrel cell (with a horizontal axis) and instead involves slantwise convection within vast horizontal Rossby waves. But the PalmenNewton model retains the ITCZ and the Hadley cell, and there are vestiges of a polar cell on the polar side of the circumpolar westerlies. The subtropical jet (STJ) lies at the edge of the Hadley cell, above the interface between descending equatorial air and cold air from higher latitudes. Similarly, the polar-front jet (PFJ) is above the interface between subpolar and subtropical air masses (Chapter 13). There is no well-defined jet above the weaker and more shallow Antarctic front.
Even the model in Figure 12.16 has the disadvantage that it is essentially static, understating the atmosphere's inherent unsteadiness. One way of demonstrating its dynamic character is to play a movie loop of the observed wind, temperature and moisture contents at thirty levels, say, and points about 70 km apart horizontally, with data for each twelve hours over a decade. This has provided detailed estimates of the meridional movements of energy and moisture within the Hadley cell, and in transient and stationary eddies. It has also given insights into the characteristics of storm tracks and teleconnections.
Alternatively, we simulate the global circulation on a large computer, creating a General Circulation Model (GCM). GCMs were first developed by Newton Phillips in 1956 and Joseph Smagorinsky in 1963, on lines pioneered by Lewis Richardson in 1922, before suitable computers were available. A GCM ignores concepts like circulation cells, jet streams and large-scale winds. Instead, it involves the physics of atmospheric processes using fundamental equations of motion (Chapter 15) and derives the temperature
change, for instance, of each unit volume of the atmosphere from the advection of heat into the volume, the net radiation input and any release of latent heat there. Likewise for changes of wind speed or moisture content during any time-step. The calculations are repeated in each successive time-step for heat, motion and moisture in each of thousands of unit volumes, starting from a description of the initial conditions (e.g. those of today) and carrying on to deduce the situation hours, days, months or years ahead. The trillions of calculations involved have become practicable only since the recent advent of very large, very fast computers. Unfortunately, the atmosphere is still represented by data from points 100 km apart, for example, so that smaller features like clouds are ignored, except indirectly. Nevertheless, GCMs are the basis of modern weather prediction, leading to clear improvements in the accuracy of forecasting
(Chapter 15). They are also essential tools in estimating climate change. This requires allowing for SST fluctuations, which can be determined by a separate model of ocean movements. Preferably, both the atmospheric and ocean models are combined (i.e. 'coupled'), and this is done in studies of the ENSO phenomenon.
Was this article helpful?