Probable fix 

 accuracy during 

 survey if using 

 present predic- 

 tions: 

 1-1/2 n. mi. 



Accuracy of 



implemented 



system: 



Day 0.7 n.mi. 



Night 0.8 n.mi. 



Transition 1 .1 n.mi. 



Overall 0.9 n.mi. 



substantially improve when the daytime propagation coeffi- 

 cients have been revalued. * 



The overall accuracy which would be actually ob- 

 tained using present prediction in the entire North Atlantic 

 area may be estimated from figures 1 and 2 and the overall 

 rms discrepancy. Apparently, a tjrpical circular error 

 probable on the order of 1-1/2 n.mi, is appropriate for the 

 24-hour day. However, results during the day would be 

 better than nominal while results at night and during transi- 

 tions would be worse; and, of course, the expected accuracy 

 varies considerably with position. In practice, actual re- 

 sults might be either better or worse. The earlier skywave 

 corrections provided to the ships were not as refined as 

 those used in the present analysis. Computations for 

 Criggion-Forestport were in error. At best, the spatial 

 density of corrections was rather sparse, thus presenting 

 a possibility for interpolation errors. However, adjustment 

 of the computed diurnals by local monitoring in port may 

 have considerably reduced the effect of prediction errors. 



The results may also be used to evaluate the accu- 

 racy available in an implemented Omega system. The appli- 

 cation is not direct but depends on assumptions of eventual 

 system geometry, possible effects of spatial correlation, 

 and the effect of operating in the absolute mode instead of 

 operating with some stations actually synchronizing as 

 slaves to a master station. It has been shown that an approxi- 

 mate conversion from LOP error in the synchronized mode 

 to fix error in an implemented system may be made by mul- 

 tiplying the present results by 0. 1 n. mi. /cec. ** Hence, the 



*See footnote on page 8 of reference 7 and also reference 4. 



**See references 2, 10, and 11. In particular, multiplica- 

 tion by 0. 106 n.mi. /cec yields rms fix error while multi- 

 plication by 0.093 n.mi. /cec yields the c.e.p. 



25 



