were at a minimum at 0600 GMT. After analysis, the data for the period when 

 the wind was blowing across the ship toward the boom were divided into 1400- 

 1959 and 2000-1359 GMT subgroups, and the other data into 2200-1259 and 1300- 

 2159 GMT subgroups. 



Figure 20 was used to obtain the lag coefficients for determining the 

 independent sample number. 



The sample means and variances for the distribution of observed tempera- 

 tures used in each comparison are given in table 7. The daytime shipboard 

 and boom temperatures show very small variances, an indication of reasonably 

 uniform conditions. The rawinsonde adjusted temperatures average almost 0.35''C 

 cooler than the shipboard temperatures and only slightly cooler than the boom 

 values. However, because of the large variation in the rawinsonde tempera- 

 tures most of the F tests applied to daytime data reject the hypothesis that 

 the shipboard- rawinsonde and boom-rawinsonde sensors were measuring the same 

 environment (see table 8) . The distribution of the rawinsonde temperatures 

 is not normal, and the emphasis in this discussion is therefore placed on the 

 K-S test, which rejects the null hypothesis for the shipboard-boom comparisons 

 when the wind is not blowing from the ship toward the boom and for the 2000- 

 1359 GMT comparison when the wind is blowing from the ship toward the boom. 

 Figure 21 presents in graphical form the results of the K-S test. This test 

 also rejects the null hypothesis for the shipboard-rawinsonde 2200-1259 GMT 

 subgroup when the wind is not blowing from the ship towards the boom, and both 

 shipboard-rawinsonde groups when the wind is blowing from the ship toward the 

 boom. Therefore, it appears that the boom measurements are the most represen- 

 tative of the environment sampled by the rawinsondes, except for the 1300-2159 

 GMT subgroup with the wind not blowing across the ship toward the boom. When 

 applied to this group, the K-S test does not reject the hypothesis that the 

 shipboard-rawinsonde sensors are measuring the same environment, and hence it 

 is of little consequence which data are used with, the rawinsonde measurements. 



5.5 USCGC Roakawccy 



The Coast Guard cutter Roakaway has a much lower profile than the four 

 Coast and Geodetic Survey ships. Because its main deck is only 19 ft above 

 the ocean surface, the boom was mounted on a platform 4 ft above the main deck, 

 placing it approximately at the same height as the area from which shipboard 

 observations were made. This area was located aft of the exhaust stacks near 

 the rawinsonde shelter some 280 ft from the ship's bow. 



Figure 22, showing the average differences between the shipboard and boom 

 temperatures, indicates that the shipboard temperatures were generally warmer 

 when the wind was not blowing across the ship toward the boom and cooler when 

 the wind was blowing across the ship toward the boom. The reason for this is 

 probably twofold: the addition of heated exhaust gases into (a) the observer's 

 environment on the ship and (b) the boom environment. The data for the period 

 when the wind was not blowing across the ship toward the boom were subdivided 

 into 0800-2159 and 2200-0759 GMT subgroups; the other data, because they repre- 

 sent only a few observations, were not subdivided. 



10 



