SECT. 2] LARGE-SCALE INTERACTIONS 255 



the Japan Sea is an enclosed basin, nearly surrounded by surface and upper air 

 stations, and regularly combed by vessels of the Japanese fisheries. 



The main physical features of the air-flow and its modification in its passage 

 across the sea, from Russia and Korea, to the west coast of Japan, are shown 

 in Figs. 73-75. As in the trade case, the inversion base rises markedly from the 

 upstream to the downstream end of the section, in correlation with the develop- 

 ment of convection which has become intense off the west coast of Japan. Also 

 similar to the trades, convective clouds are most vigorous where the wind speed 

 and sea-air temperature differences are largest. As seen in Fig. 75, the down- 

 stream increase in potential temperature suggests a large oceanic heat source ; 

 however, a careful budget study including all terms in the heat balance is 

 needed to separate this input from precipitation and compressional warming. 



Manabe performed a heat and moisture budget for a box the bottom area 

 of which was the Japan Sea polygon shown in Fig. 72 extending vertically to 

 500 mb. Its surface area is 0.76 x lO^^ cm^, or about one-third of the Caribbean 

 ellipse of Colon. Here equations (27b) and (28b) had to be formulated in a 

 manner to include the important effects of time dependence and were actually 

 set up for the computation in the following forms : 



LP + Qs + Ra = dhldt = lti\ \ Cn{CpT + Agz) dl{dplg) 



+ Aj.s.[wtpt{cpT + Agz)t - Wbpb{CpT + Agz)b] (27e) 

 and 



Qe-LP = dLqIdt = 2ti CnLq dl{dplg) + Aj.s.[wt'pt{Lq)t-Wbpb{Lq)b], (28e) 



where the bar denotes time averaging over the period, the subscript J.S. 

 denotes the Japan Sea polygon, and the symbol 2« denotes time integration 

 in a series of finite steps. The subscript t denotes top and the subscript b denotes 

 bottom. 



The following assumptions were made : 



(i) Local time-dependent terms, d/dt, in both potential temperature and 

 mixing ratio were neglected compared to advective changes and sources and 

 sinks. 



(ii) Radiation, precipitation and vertical advection were computed entirely 

 from soundings and data averaged over the whole fifteen day period, so that 

 fluctuations and vertical eddy transports were left out. Turbulent fluxes 

 through the 500 mb surface were set to zero on the grounds that the inversion 

 base was always below this. 



(iii) The velocities in the horizontal advective terms were evaluated from 

 the geostrophic approximation, using the contour heights of pressure surfaces 

 with a grid containing six divisions on either side of the Japan Sea. At the 

 surface, the horizontal advection was reduced to one-half the geostrophic in 

 rough correspondence to the observed ratio of actual to geostrophic winds at 



