Of the 66 profiles, 20 were made when the XBT systems malfunctioned. The statistics 

 for all the differences for each depth were very similar to the results obtained from the 

 first two studies. The major difference was that the standard deviations were somewhat 

 larger for this study. The statistics for the differences associated with the profiles made 

 when the systems were not malfunctioning were also similar to the statistics for the first 

 two studies. 



Simultaneous XBT and thermistor chain measurements were made with the vessel 

 both hove to and underway at 3 knots. Comparisons were made at 44 depths spaced 

 from the surface to 242 m, the maximum depth of the thermistor chain. From a study of 

 these measurements, it was concluded that the probes used measured temperatures from the 

 surface to 242 m that were slightly less than those measured by the thermistor chain. The 

 biasing of these differences may be related to using a "run" of eight consecutive probes. 

 While underway at 3 knots, 73 XBT profiles were attempted. Comparisons of the 64 

 successful and partially successful profiles with single thermistor chain temperature scans 

 that started at the same time the XBT probe was released showed that three profiles 

 were grossly in error and these were eliminated from the data set. Analysis of the remaining 

 61 profiles suggested that below the near-surface layer, the thermistor chain measured 

 temperatures slightly higher than the XBT measurements. This possibly resulted from a 

 slight shoaling of the thermistor sensors as a result of the 3-knot towing speed. Since the 

 thermistor chain depth sensor was inoperative, it was not possible to check this conclusion. 

 Comparisons of the XBT measurements with thermistor chain measurements made in the 

 near-surface layer showed that the 61 XBT profiles measured temperatures slightly higher 

 than the thermistor chain. The average of 1037 comparisons was 0.04°C with a standard 

 deviation of 0.13°C; 62.8% of the differences were positive; 35.3% were negative; and 1.9% 

 were zero. These results are consistent with those obtained from the hydrocast and STD/SV 

 as well as the quasisimultaneous STD/SV comparisons. In addition, a serial tabulation of 

 the average differences, standard deviations, and the number of positive and negative 

 differences for each of the 6 1 XBT profiles showed temperature measurement biases both 

 with respect to measurements made by individual XBT probes and with respect to "runs" 

 of consecutive probes. 



PROFILES WITH LARGE DIFFERENCES 



The analyses summarized above were concerned primarily with average temperature 

 differences. The results suggested that, on the average, the XBT system measured temper- 

 atures in the surface-to-400-m depth interval that were slightly higher, and vertical 

 temperature gradients that were slightly larger, than those measured by hydrocast, STD/SV, 

 and thermistor chain systems. In many applications, the concern is not with average values 

 but rather with the use and interpretation of individual, or a short series of individual, 

 profiles. A set of profiles that reached a minimum depth of 200 m and whose temperature 

 exceeded the average hydrocast and STD/SV temperatures by more than ±0.50°C at 200, 

 300, or 400 m was compiled. This set contained 54 profiles, which was 9.7% of the total 

 set of 559 profiles. This set does not include the profiles made with the malfunctioning 

 XBT systems. An examination of these data showed positive differences as large as 3.02°C, 

 negative differences as large as -0.73°C, positive temperature gradient biases as large as 

 0.96°C/I00 m, and negative temperature gradient biases as large as -0.73°C/100 m. 



129 



