SECT. 2] LARGE-SCALE INTERACTIONS 129 



we must then be able to break down the total transports and analyze the in- 

 dividual components in air and sea. The sea-transport divergence, Qvo, has 

 already been obtained as residual from equation (1) by our previous heat 

 balance for the earth's surface, for which evaporation distributions were pre- 

 viously evaluated. Thus, to complete the heat budget in (25) we need to know 

 the distribution of precipitation, P, and the flux divergence of sensible heat and 

 potential energy in the atmosphere, Qia. 



The procedure here (and in Table I) is to start with the radiation balance, 

 Rs, of Bagrov, then to present the best available precipitation figures and their 

 latitudinal dependence, and finally to arrive at Qia as residual in (25) to com- 

 pare with more direct meteorological determinations of the various atmospheric 

 transports. We are thus following most closely the Russian development 

 (Budyko, 1956) and presenting comparisons of their results with the more 

 accessible western calculations where possible. It is a considerable credit to the 

 recent advances in both meteorology and oceanography that the various 

 independent pieces of the puzzle fit together so consistently. The average 

 annual global energy transactions are now fairly well-known ; they both give 

 clues to building dynamic models, and specify constraints that these models 

 must satisfy. 



b. Mean annual distribution of precipitation over the oceans 



Adequate assessment of oceanic rainfall poses a problem. Direct shipboard 

 measurements are very difficult to obtain reliably, and therefore extrapolation 

 from island and coastal stations must generally be undertaken. Thus, in addi- 

 tion to uncertainty whether coastal rainfall is a fair measure of that over the 

 open sea, wide portions of the mid-ocean regions remain insufficiently sampled. 

 However, since World War II, meteorological studies of oceanic and island 

 precipitation statistics and dynamics have advanced considerably, particularly 

 in the tropics ; they provide both the opportunity of spot-checking the extra- 

 polated patterns and of testing whether such extrapolation is physically 

 compatible with rainfall dynamics. 



Fig. 14 presents the most recent global picture of oceanic rainfall patterns 

 (Drozdov, 1953) which was obtained by direct extrapolation between stations 

 without any arbitrary reductions to correct for "land effect". We may compare 

 this figure with the presentation of Jacobs, based on results by Wiist (1936). 

 These were reduced in places by 20-30%, to agree with then existing evapora- 

 tion figures, the required reduction being explained as a correction for presumed 

 coastal enhancement of rainfall. The isohyetal distributions in Fig. 14, how- 

 ever, are almost entirely similar to those of Jacobs, with patterns and maxima 

 which are entirely superposable. A numerical comparison is made in Table 

 VIII ; Drozdov's amounts are, as expected, almost everywhere larger than 

 those of Jacobs and Wiist, by an average of about 20%, in better agreement 

 with still earlier determinations by Meinardus (1934). Overall budget studies 

 (Riehl and Malkus, 1958) suggest that the higher Russian values are the better 



