210 MALKUS [chap. 4 



suggests, but does not prove, downward heat flux, since counter-gradient heat 

 flows are not unusual in the atmosphere (Bunker, 1956). Much more important 

 mechanisticaUy, however, it shows that the small turbulent eddies are com- 

 monly "overshooting" their level of zero buoyancy at least a thousand feet 

 below cloud base, so that wind stirring is necessary to convey the water vapor 

 the remaining distance to the condensation level. This is but one link in the 

 now firm chain of evidence that the oceanic trade cumuli, unlike their con- 

 tinental relatives, do not have individual "roots" in buoyant cloud-scale 

 thermals penetrating up into them from the sea surface. In fact, the aircraft 

 records show that small-scale turbulence is no more developed on flights just 

 below cloud bases than on those in the intervening clear spaces. 



b. The origin of trade-wind cumuli 



The evidence that small-scale convective turbulence near the sea surface 

 and the cumulus activity in the cloud layer are somewhat decoupled from each 

 other was an unexpected result of the early Woods Hole expeditions. It raised a 

 serious puzzle about the origin of the trade cumuli, the important role of which 

 in the overall circulation we have brought out in the preceding sections. 



One of the most striking and suggestive features of the trade-wind clouds is 

 their arrangement into groups, separated by comparable or somewhat larger 

 clear spaces. Frequently, especially in the wet season, these groups consist of 

 50-100 km long lines oriented parallel to the flow, while on other occasions the 

 bunching appears to be purely random (Riehl, Gray, Malkus and Ronne, 

 1959). The orientation into lines becomes more pronounced as the conditions 

 for convection are more favorable, particularly when synoptic-scale con- 

 vergence in the wind field is present, while the random bunching is more 

 characteristic of inversion-dominated situations and the outer fringes of the 

 trades. In both cases, the cloud groups break out in those regions where the 

 homogeneous layer is thickened relative to its surroundings and the slightly 

 stable "transition layer" of Fig. 52 is absent. Thus the cloud base is, within 

 observational error, at the level of water-vapor condensation of the average 

 sub-cloud air. Its height is readily calculated from a tephigram (or other 

 meteorological thermodynamic diagram) by carrying low-level air upward 

 along a dry adiabat to saturation. 



The depth of the homogeneous layer averages about 550 m, with variations 

 of about 20% in space and as much as 100% in time. Extreme day-to-day 

 ranges are about 300-800 m. The space scale of the depth variations suggest 

 their connection with the longer-period "eddies" of Figs. 51a and 53, which 

 appear to retain their phase throughout the mixed layer. Thus there is evidence 

 of horizontal convergence and divergence on the cloud-group scale, so that it is 

 still possible to seek the "roots" of the cumuli in deformations of the sub-cloud 

 layer as a whole. When these take the form of long rolls oriented with the wind, 

 some kind of instability to this type of convective motion is suggested, although 

 it is not clear whether the instability is dynamic, thermal or some combination. 



