Table 1 shows the sample means and variances for the distributions of 

 the observed temperatures used in each comparison. In both cases, the ship- 

 board temperature averages warmer than the boom and rawinsonde values in the 

 group that represent primarily daytime observations and cooler in the night- 

 time group. The maximum difference is in the 1300-0359 GMT group, where the 

 shipboard temperature averages 0.37°C warmer than the boom and 0.29°C 

 warmer than the rawinsonde temperatures. When the wind is blowing across 

 the ship toward the boom, the boom temperatures apparently tend to be more 

 stable at night, as shown by the smaller standard deviation. 



Figure 9 was used to determine the lag coefficients that were applied 

 to each sample to obtain the independent sample size and effective degrees 

 of freedom. 



Results of the parametric F^ Student's t, and the nonparametric K-S test 

 are shown in table 2. When the wind is blowing across the ship toward the 

 boom, the hypothesis that the boom, shipboard, and rawinsonde sensors are 

 measuring the same environment is not rejected, and it would be difficult to 

 say whether the boom or shipboard data would be better to use for comparison 

 with the rawinsondes . When the wind is not blowing across the ship toward 

 the boom, however, the hypothesis that the shipboard and boom sensors and 

 the shipboard and rawinsonde sensors are measuring the same environment is 

 rejected both by the Student's t test and the K-S test for the period 1300 to 

 0359 GMT. The temperature differences referred to earlier are significant. 

 Comparison of the boom and rawinsonde data, however, does not reject the 

 hypothesis according to the results of the tests discussed here. 



In summary, the boom data are probably the better measurement of the 

 temperature environment being sampled by the rawinsonde during the day, par- 

 ticularly when the wind is not blowing across the ship toward the boom. 

 The 0400-1259 GMT subgroup, as in the case when the wind is blowing across 

 the ship toward the boom, shows that the hypothesis is not rejected. Which 

 ambient temperatures would be better to use with the rawinsondes is open to 

 question; the ship's warmth appears to add 0.1°C to 0.2°C to the air passing 

 over the ship. 



From table 2, the following assumptions can be made: 



(1) The boom temperature is the better one to use with the rawinsondes 

 when the wind does not blow across the ship toward the boom. 



(2) The ship's and boom's temperatures are the same in the periods 

 1000-2159 and 2200-0959 GMT when the wind blows from the ship 

 toward the boom but are not the same in the periods 1300-0359 

 and 0400-1259 GMT when the wind does not blow across the ship 

 toward the boom. 



Figure 10 shows the cumulative probability curves for the shipboard-boom 

 shipboard- rawinsonde, and boom-rawinsonde (°C) temperature comparisons for 

 the 1300-0359 GMT subgroup of data for the case when the wind is not blowing 

 across the ship toward the boom. The K-S maximum D value and the calculated 

 D(, values are indicated, based on data obtained between June 19 and July 2, 

 1969. 



