98 The Three-dimensional Temperature Distribution and its Variation in Time 



radiation. This controversy was settled by important and interesting experiments in 

 the sense of Nansen's reasoning. He suggested that the winter convection in parts of 

 the Norwegian Sea and of the North Atlantic (south and south-east of Greenland and 

 in the Irminger Sea) could reach very great depths because of the almost uniform den- 

 sity structure of the sea, so that the autumn and winter cooling thus continued almost 

 to the bottom. This should therefore be the place where the uniform North Atlantic 

 Deep Water was formed. The observations of the winter cruises of the "Meteor" 

 in the Iceland-Greenland waters during 1929-35 have shown that these views of 

 Nansen were correct. The cause of this convection is certainly the radiation of the 

 surface layer during the late autumn and early winter. 



In the North Polar Basin conditions are somewhat different. The very large rivers 

 of Asia and North America bring large amounts of fresh water into this basin and these 

 overlay the saline water that flows into the deeper layers from the Atlantic Ocean. 

 Any deep-reaching convection is scarcely possible here, cooling is limited to the sur- 

 face layer and is correspondingly stronger. The melting of ice in the spring and summer 

 sets up a barrier against the denser water masses in the deeper layers so that the sum- 

 mer heating does not penetrate far. 



Characteristic examples of a convection that extends to great depths, and can be 

 attributed primarily to an increase in the salinity of the surface layer caused by strong 

 evaporation, are found in the Mediterranean Sea and in the Red Sea. The low precipi- 

 tation, the small amount of river water flowing in, and the high rate of evaporation 

 raise the salinity of the surface layers especially in the summer, though at this time only 

 a limited convection occurs, since the increase in density is largely offset by the effect 

 of the summer heating. However, in the autumn and winter a well-developed con- 

 vection is set up due to the lowering of the temperature of the surface water and 

 reaches to great depths because of the uniformity of the vertical structure of the deeper 

 layers. 



The accurate mathematical treatment of thermo-haline convection processes is not 

 easy. It can be attempted in the following way (Defant, 1949). To begin one considers 

 two thin layers of thickness h-^ and h^, temperature ^i and d'z, salinity ^i and S^ and 

 density pi and p^. A disturbance introduced in the entire upper layer so that p^ = p^ 

 will cause mixing of the two layers /zj and h^ by convection and the final result will 

 be the layer h^ + h^ of density p^. If the disturbance in the upper layer is assumed to be 

 due entirely to a reduction in the temperature of the upper layer by '&-y — Ad'i then the 

 final temperature at the end of convection results to 



(^, - A{^,)h, + ^Jh h, 



— h^^. — ^^--^:;^^- 



when the mean temperature that would be obtained by simple mixing of the initial 

 water masses is given by 



^'•'^ h, + h, ' 

 The final salinity after ceasing of convection is given by 



Syh^ + S^h^ 



^'''~ h, + h, 

 and corresponds to the salinity obtained on simple mixing. 



