Two comments on the data presented in table 27 and figure 26 are indicated: 



a. Once the profiles made when the XBT systems were malfunctioning 

 were detected and removed from the data set, the XBT and indepen- 

 dently measured surface temperatures were in agreement. The average 

 of 467 comparisons was 0.02°C, with a standard deviation of 0.15°C. 

 The distribution of the differences was slightly biased, with 56.5% being 

 positive, 38.6% negative, and 4.9% zero. 



b. All the XBT profiles used in this analysis were originally classified as 

 visually acceptable profiles. In the absence of the independently mea- 

 sured surface temperatures, all of the 736 XBT surface temperatures 

 would have been considered valid. In that event, the statistics are 

 markedly altered, and the average difference is -0.1 5°C with a standard 

 deviation of 0.47°C. In addition, the bias in the distribution of differ- 

 ence reverses, with 56.3% of the differences being negative, 40.3% 

 positive, and 3.4% zero. 



Finally, it is concluded from this analysis that a method of detecting malfunctioning 

 XBT systems that produce visually acceptable XBT profiles is required. 



TEMPERATURE ACCURACY OF 460-m SYSTEMS 



This section contains the results of four studies that compare XBT measured 

 temperatures at selected depths with average hydrocast and STD/SV measurements, quasi- 

 simultaneous STD/SV measurements, and simultaneous thermistor chain measurements. 

 The profiles whose calibration lines deviated from 16.7°C by more than ±0.34°C are not 

 included in these studies. 



COMPARISON WITH AVERAGE HYDROCAST 

 AND STD/SV TEMPERATURES 



The results of two studies that compared XBT measured temperatures with hydrocast 

 and STD/SV measurements are discussed in this section. These studies included all visually 

 acceptable XBT profiles made during the Gulf of Alaska experiments in water mass 2 and the 

 transition water mass; during SUDS I 1972 in area C; during CAPER in water mass 2; and 

 during RAPLOC/DEEPTOW when the ORB was at anchor. In these water masses, enough 

 hydrocast and/or STD/SV measurements were made in the area and time interval over which 

 the XBT profiles were themselves made to provide a reasonably accurate average temperature 

 and an estimate of the standard deviation. The first study includes all visually acceptable 

 profiles reaching a minimum depth of 200 m, and the second all visually acceptable profiles 

 reaching 400 m and also having an independent surface temperature measurement. 



Visually Acceptable Profiles Reaching 200 m 



Comparisons were made at 200, 300, and 400 m. These depths were selected 

 because they are below the thermocline where vertical temperature gradients and spatial and 

 temporal variations are small. The analysis began with the computation of the differences 

 between the XBT 200-, 300-, and 400-m temperatures and the appropriate average hydro- 

 cast and STD/SV temperatures listed in table 13. 



The most accurate and the most extensive of the five sets of differences was 

 obtained during the Gulf of Alaska experiments in water mass 2, where 65 STD/SV profiles 

 were made in the same area and time interval over which 270 XBT profiles were made. In 

 water mass 2, the 200-, 300-, and 400-m temperatures were extremely stable. The standard 

 deviation of the average hydrocast and STD/SV measurements at these three depths was 

 0.04°C (see table 13). The second most accurate set of differences was obtained during the 



63 



