SECT. 2] LARGE-SCALE INTERACTIONS 167 



extremes, so that its position and migration are determined largely by the 

 distribution of heat sources and sinks, in particular by the strength of the cold 

 source at high latitudes in the winter hemisphere. There thus does not appear 

 to be any geometrically fixed equatorial boundary condition for the atmos- 

 phere's general circulation. 



For the determinations of exchange to enter as source terms in (27b) and 

 (28b), we should like an independent estimate of Qs and Qe from the transfer 

 formulas and an energy budget check on their sum, at least, from (lb). Un- 

 fortunately this is not possible because data to assess Qs from the transfer 

 formula are lacking. For Qe, equation (21), the Atlas of Climatological Charts 

 of the Oceans (U.S. Weather Bureau, 1938) and Bean and Abbot's (1957) 

 surface humidity maps were used to derive Table XIII into which contribu- 

 tions from land areas have been weighted. For the continents of the humid 

 tropics, the same evaporation rate was used as over the oceans ; over deserts 

 it was taken as zero. These assumptions are well justified by the continental 

 evaporation figures reported by Budyko (1956). Using the results for Qe we 

 may find Qs as residual in (lb) after entering London's figures for Rs and Ra, 

 and provided some estimate of Qvo may be made. Riehl and Malkus assumed 

 that it would be a negligible term in the ocean's heat budget. In equatorial 

 regions where sea-temperature gradients are slight and currents zonal, it was 

 envisaged as even smaller than in the Caribbean, where Qvo was within the 

 computational uncertainty of the larger balance components. 



In the following analysis, we consider mainly the belt extending from the 

 troughline to a distance of 10° latitude on the winter side. In this belt the 

 solution of (lb) for Qs (neglecting Qvo) gives 



Qs = 0.3l{Rs) + l.00{- Ra)-0.dS{Qe) = 0.38 units (32) 



which implies a Bowen ratio {QsjQe) of 41% and is much larger than expected 

 from other determinations in the tropics. If, as suggested by Fig. 13, Qvo in 

 these regions is as much as 20% Qe, we obtain a Qs of 19 units and a more 

 plausible Bowen ratio of 20%. Although 19 units are within the accuracy of the 

 radiation computations, a 20% reduction in London's R— Rs — Ra is less 

 likely in view of the agreement therein with Budyko (table VI, page 121). 

 Furthermore the Qe's in Table XIII are in good agreement with all recent 

 evaporation determinations i for the region and 25% greater than that of Wiist 

 (1936) ; the Qe in (32) can hardly be much underestimated. We shall, therefore, 

 accept a larger ratio oi QsjQe in this zone than in the trades (3-10%) as a possi- 

 bility to be explored, particularly since a plausible mechanism exists to explain 

 it, as we shall see. 



1 To compare Qe in Table XIII with the results in Table XI we must recall that Colon 

 considered a geographically fixed locality at 10^-20°N, which is, relative to the trough, 

 roughly 0^-10" in summer and 10^-20^ in winter. 



