SECT. 2] 



liABGE-SCALE INTERACTIONS 



185 



100° 90° 80° 70° 60° 50° 40° 30° 20° 10° 



ATLANTIC 

 MEAN WIND STRESS 



20° 30* 



Fig. 43. Mean wind stress for July over the North Atlantic computed from transfer formula 

 (17). (After Scripps Institution of Oceanography, Oceanographic Report No. 21, 1950.) 



frequency of winds in sixteen compass directions by month and by 5-degree 

 quadrangles. The procedure here was to compute the stress from (17) and the 

 mean wind speed in each Beaufort interval, using cz)= 1.7 x 10~3 for Beaufort 

 force 4, where the mean speed is 6.7 m/sec, or the transition of the dashed 

 curve in Fig. 6. The Atlantic method was, therefore, cruder than that used in 

 the Pacific, especially since the data required that sometimes two Beaufort 

 intervals had to be lumped together. The results for the mean annual picture 

 are shown in Fig. 41, while Figs. 42 and 43 reproduce their Atlantic charts for 

 January and July, respectively. 



b. Range of validity and representativeness of the computed stress distributions 



We have made numerous tests for the purpose of evaluating the computa- 

 tional methods of the Scripps reports and the representativeness of their 

 results, within the framework that equation (17) is basically a correct formula- 

 tion. It should be kept in mind that this is only sound for the tropics, while in 

 temperate latitudes any such stress determinations may be off by a factor of 

 two due to thermal stratification. What we have attempted to analyze is the 

 effect of assumed variation in Cd, the frequency distribution of wind speeds and 

 the chances that, in any given month, the vector wind stress computed via (17) 

 from actual ship measurements would resemble the climatic mean stress for 

 that month. We have taken one location typical of the tropical Atlantic (just 

 north of San Juan, Puerto Rico, at latitude 19° 30'N, longitude 66° W) and one 



