SECT. 2] LARGE-SCALE INTERACTIONS 175 



suffices for balance up to tlie average top of the moist layer. The latter is itself 

 determined by the altitude to which normal entraining cumuli, giving off their 

 heat and moisture during ascent, can penetrate. In fact, it is known from 

 cloud observations that the mean top of the cumuli increases from the base of 

 the inversion (6-8000 ft) in the outer trades, to about 12,000 ft in the equa- 

 torial trough zone, in fair agreement with the level of minimum Q in Fig. 36. 



The reason for the upward decrease of the effectiveness of diffusion lies in 

 the logarithmic dependence upon temperature of the air's capacity to hold 

 moisture. In spite of all variations in individual soundings, the broad course of 

 atmospheric specific humidity in the vertical closely parallels the logarithmic 

 decay curve. At 500 mb, for example, the average moisture content of the air 

 is an order of magnitude down relative to the sub-cloud layer. Thus in the 

 total Q transport, the upward diffusive flux of moisture becomes overbalanced 

 by downward flux of sensible heat. Since diffusive warming of the equatorial 

 troposphere from stratospheric levels can be precluded, it follows that the 

 transport comes from the surface, against the gradient, by some kind of 

 selective mechanism which also produces the necessary net mass ascent to 

 complete the meridional cell. 



Going downward in the motion scale, one might visualize that the energy 

 transport takes place in more restricted areas of ascent connected with tropical 

 synoptic disturbances, such as the easterly wave or equatorial vortex (see 

 Riehl, 1954). These occupy at most one-tenth of the area and hence would not 

 enter the climatic mean sounding of Fig. 36 with any weight. A deep moist 

 layer is built up by convergence in these disturbances and, conceivably, the Q 

 minimum could thereby be eliminated. This supposition fails to meet the test 

 of radiosonde data which show a mid-tropospheric Q minimum in the rain areas 

 of even intense tropical storms. 



This dilemma led to a revolutionary concept of vertical transfer mechanisms 

 in the tropics, the so-called "hot tower hypothesis" of Riehl and Malkus. It 

 points to the giant cumulonimbus chimneys of equatorial disturbances (Fig. 3) 

 as the mechanism of ascent, vapor combustion and upward energy pumping. 

 The rising portion of the meridional cell is concentrated in the restricted regions 

 of the towering cloud bands within the tropical storms and operates by means 

 of selective buoyancy on the large-cumulus space and time scale ! 



In the ordinary, small-sized trade cumuli, buoyancy is wasted away by 

 mixing with the surroundings ; few of their tops penetrate more than half a 

 kilometer into the dry air above the inversion. But within the general rain 

 area of equatorial disturbances, there are imbedded central cores in the penetra- 

 tive giant cumulonimbus which are protected from dilution by the large cross- 

 sections of the towers. In this way, the high heat content, ocean-exposed air 

 of the sub-cloud layer can be pumped to great heights to balance the heat 

 losses and provide the calculated exports. 



Since its formulation, the hot tower hypothesis has been extended theoreti- 

 cally and observationally strengthened. Studies of large-cumulus dynamics 

 (Levine, 1959; Malkus, 1960) and measurement programs incorporating 



