COMPARISON WITH EXPERIMENTAL RESULTS 



347 



Figure 8.21 shows the results of the specific measurements reported by 

 Anway for the radio sextant for all cases at an elevation angle of 8° zt 

 0.09°; each point represents an "instantaneous" reading. The soHd hne 

 represents the Hnear regression of the measured refraction data on the 

 values of A''s; the dashed line shows the predicted linear relationship 

 derived from least squares fits to the CRPL Standard Sample ray traced 

 refraction data. The mean bias between the two lines is about 40 jurad, 

 interestingly close to, and in the same direction as, the apparent cali- 

 bration error noted in the mean refraction data at high elevation angles. 

 The standard error of estimate is considerably higher than predicted ; how- 

 ever, the rms uncertainty of ±0.052°, or ±0.91 mrad, in the apparent 

 elevation angle would be sufficient by itself to increase the standard error 

 of estimate to about ±0.017 mrad, which is 4 times larger than the pre- 

 dicted value. It is not known how much of the total standard error of 

 ±0.12 mrad is due to measurement errors as opposed to unforeseen 

 fluctuations in actual atmospheric refraction. 



< 



< 



2.5 



< 



Li_ 

 LJ 



1.5 



;LEAST SQUARES FIT 



-0.216 + 0.00754 Ns (mrad) 

 0.87, S.E. = 0.12 mrad $ Ij 



\^ 



MEASURED REFRACTION OF 1.85 cm RADIO WAVES 

 AT 8° ± 0.052° ELEVATION ANGLE , WITH 

 UNCERTAINTY WINGS DUE TO INEXACT 

 ELEVATION ANGLE : rdX.] „„ = 0.0163 



^LINEAR REGRESSION OF C.R.P.L. STANDARD 

 N- PROFILE SAMPLE; PREDICTED VALUES; 

 T = -0.058 + 0.00698 Ns (mrad) 

 r = 0.999, S.E. = 0.004 mrad 



\ \ \ \ 



290 300 



310 



320 



330 



340 



350 360 370 380 390 



Figure 8.2 L 1.85 cm radio refraction at an elevation angle of 8 degrees, at Cedar Rapids, 



Iowa. 



(After Anway, 1961). 



