46 contain eight pairs of profiles that illustrate the nature of these differences. For example, 

 one pair showed minor differences in the near-surface layer and identical agreement in the 

 thermocline. However, below the thermocline they began to differ, with the difference 

 being an an increasing function of depth. At 400 m, the difference had increased to 3.0 1°C. 

 Generally, when pairs of profiles exhibited large differences, the profile measuring the 

 higher temperature was the incorrect profile. This is important since, if undetected, the 

 temperature errors of these profiles are systematic, biasing, and nonrandom. 



Comparison of ten 1 830-m profiles with two sets of simultaneous profiles made 

 on two different 460-m systems showed that 1 5 (75.0%) pairs of the 20 pairs agreed to 

 within ±0.34°C at all above and below-thermocline comparison depths. For 17 (85.0%) 

 pairs, the layer depth difference was less than or equal to ±5 m. On the average, the 1 830-m 

 system measured temperatures that were slightly lower than the temperatures measured 

 by the two 460-m systems. 



PROFILES TRANSMITTED TO FLEET NUMERICAL 

 WEATHER CENTER 



During the SUDS I 1972 experiments, the DE STEIGUER routinely digitized and 

 transmitted XBT temperature data to the Fleet Numerical Weather Center (FNWC), 

 Monterey, CA, where they were used as inputs to predictions of propagation loss and 

 Fleet sonar performance. From 8-23 February 1972, 31 profiles were attempted, with 

 three being catastrophic failures. The remaining 28 profiles were judged to be visually 

 acceptable by the observer and were digitized and transmitted. Of the 28 profiles, 15 

 were made in area C where enough hydrocast and STD/SV measurements had been taken 

 to establish an average 20O, 300-, and 400-m temperature. Of these 15 profiles, 13 

 reached 400 m. Of the 13, the number satisfying the accuracy criteria at all three depths 

 was three (or 23.1%); at two depths was two (or 15.4%); at one depth was three (or 23.1%); 

 and at no depth was five (or 38.5%). Thus, out of 13 profiles reaching 400 m, only 23.1% 

 satisfied the accuracy criteria at all three depths, and 38.5% did not satisfy the criteria at 

 any of the three depths. Yet, all 13 profiles were deemed acceptable by the observer and 

 transmitted to FNWC as inputs to operational acoustic parameter predictions. In addition, 

 the differences varied from -0.36°C to 1 .09°C, with four (9.3%) of the 43 differences 

 greater than +0.50°C. 



CONCLUSIONS 



Evidence has been presented to support the following conclusions concerning the 

 accuracy of XBT temperature measurements: 



• XBT system errors varied from -0.1 9°C to 0.23°C. To determine these errors 

 independent and simultaneous temperature measurements are required. In the absence of 

 such measurements, the absolute value of the XBT recorded temperature is questionable to 

 approximately ±0.2°C. 



• Properly functioning XBT systems may develop malfunctions while making a 

 series of profiles and produce visually acceptable, but erroneous, temperature profiles. 

 Other information, such as independent temperature measurements, is required to detect 

 and identify such malfunctioning 460-m XBT systems. In these studies independent surface 

 temperatures were measured concurrently with 736 visually acceptable XBT profiles made 



131 



