47.8% satisfied a ±0.34°C accuracy specification at all 

 comparison depths. 



68.9% satisfied the accuracy specification at all in-layer and 

 below-thermocline depths. 



31.1% did not satisfy accuracy specifications at one or more 



depths. 



90.0% of the near-surface-layer depth differences satisfied the 

 accuracy specification for depth. 



10.0% of the near-surface-layer depth differences did not satisfy 

 the accuracy specification for depth. 



• Of a total of 1 11 6 comparisons of simultaneous pairs of XBT temperatures, the 

 absolute value of the differences exceeded two standard deviations of the manufacturer's 

 specified accuracy for 16.6 percent of the comparisons. 



• An examination of individual simultaneous pairs of visually acceptable 460-m 

 profiles showed that many of the pairs differed by large amounts. Some of the pairs 

 measured large differences starting in the near-surface layer, with the differences being a 

 variable function of depth, while others agreed identically in the near-surface layer and 

 began to differ at some depth below the thermocline, with the difference being an increasing 

 function of depth. 



• During the SUDS I experiments, 28 XBT profiles were judged by the observer to 

 be visually acceptable and were digitized and transmitted to the Fleet Numerical Weather 

 Center where they were used as inputs to predictions of Fleet sonar performance. Of the 

 28 profiles, 1 5 were made in area C, where enough hydrocast and STD/SV measurements 

 were taken to establish an average 200-, 300-, or 400-m temperature. Of the 15, 13 reached 

 a depth of 400 m. Of those reaching 400 m, the percentages satisfying the 200-, 300-, and 

 400-m accuracy criteria at the various depths were as follows: 



All three depths: 3, or 23.1% 



Two depths: 2, or 15.4% 



One depth: 3, or 23.1% 



No depths: 5, or 38.5%. 



If this sample of 13 XBT profiles is representative of the data being transmitted to FNWC 

 on a routine basis, the inclusion of many visually acceptable but actually erroneous 

 measurements must certainly have an adverse effect on the accuracy of the acoustic pre- 

 dictions based on these measurements as inputs to the predictions. 



RECOMMENDATIONS 



• As a result of its review of these studies, the Sippican Corp suggests that the 

 following procedure be used for those appHcations requiring retention of full available 

 system accuracy. 



". . . cahbrate with an A2A test canister whenever (1) a new 

 roll of chart paper is installed, (2) at four-hour intervals during 

 continuing drops, and (3) whenever the 2-second, mid-scale 



133 



