22 A WORLD ATLAS OF ATMOSPHERIC RADIO REFRACTIVITY 



Table 3. Approximate total standard errors, or, for N(z) 

 maps in appendix A. 



Average 



Minimum 



Maximum 



3 km 



As percent of mean N(z) 



1.0 

 0.6 

 1.3 



It is probable that the percent errors given in 

 table 3 do not increase materially above 16 km ; 

 asymptotic values of 3, 2, and 4 percent for 

 average, minimum, and maximum relative 

 standard errors are probably good estimates for 

 altitudes from 20 to 30 km, while above 35 km 

 the standard errors are probably no more than 

 0.2 A7-units. 



There are also undoubtedly some bias errors 

 involved in the N(z) maps, although they are 

 probably quite small. Sources of these bias er- 

 rors would include the equation for A7 itself 

 (Smith-Weintraub formula), the radiosonde in- 



-WEATHER SUMMARY DATA 



Figure 8. Correlation of AN; monthly mean N-profiles versus monthly 

 mean weather summary data. 



strument (where the sensor lag would seem to 

 assure a slight positive bias for all upper-air A7 

 values, even over long-term means) , and certain 

 peculiarities in the mapping and curve-fitting 

 procedures which might produce bias for some 

 values of z and not others (e.g., the bias due to 

 the sharp "knee" of the D x and D 2 exponential 

 terms at the A/-tropopause as compared with 



the smooth transition of real mean iV-profiles; 

 note example in fig. 2). It should be noted that 

 this last type of bias error was (automatically) 

 included in the mean absolute errors given in 

 table 2, since there was no easy way of separat- 

 ing this type of error from the others. Hence 

 table 3 includes an allowance for this particular 

 source of bias error. 



