PRESENTATION OF N DATA 15 



meteorological data gathered by the Canterbury Project [23] in their 

 intensive study of ranges of over-water radar signals. Both B and M 

 units assume a standard atmosphere with a linear decrease of A'^ with 

 height and thus introduce a correction which increases linearly with 

 height. If the actual atmosphere had in fact the assumed A^ distribution 

 then, for example, B{h) = B{0). Recent studies [24, 25, 26] have shown 

 than an exponential decrease of A'^ with height is a more realistic model of 

 the true atmosphere. 



For example, the data given on figure 3.3 were chosen to represent the 

 extremes of average N profile conditions over the United States. The 

 Miami, Fla., profile is typical of warm, humid, sea-level stations that 

 tend to have maximum refraction effects while the Portland, Me., profile 

 is associated with nearly minimum sea-level refraction conditions. Al- 

 though Ely, Nev., has a much smaller surface A^ value than either Miami 

 or Portland, its A^ profile falls within the limits of the maximum and 

 minimum sea-level profiles. The A'^ distribution for the 4/3 effective 

 earth's radius atmosphere is also shown on figure 3.3. 



It is quite evident that the 4/3 earth distribution has about the correct 

 slope in the first kilometer above the earth's surface but decreases much 

 too rapidly above that height. It is also seen that the observed refrac- 

 tivity distribution is more nearly an exponential function of height than a 

 linear function as assumed by the effective earth's radius model. The 

 exponential decrease of N with height is sufficiently regular as to permit a 

 first approximation of average vV structure from surface conditions alone. 

 Consider that 



N{h) = Nsexp i-h/H), (1.34) 



where H iasi scale height appropriate to the value of A'' at zero height, Ns- 

 Average values of N^ and H for the United States are approximately 313 

 and 7 km respectively. As used here, scale height is simply that height 

 at which N{h) is l/e of A''^ under the assumption of (1.34). 

 One would exj)ect the gradient 



^^ = - (Ns/H) exp i-h/H) (1.35) 



also to be sufficiently well-behaved as to allow prediction of at least its 

 general features. In fact a high (;orrelation between AN, the simple 

 difference between A" at the earth's surface and at 1 km above the earth's 

 surface, and A^,, has been observed in a number of regions. This is 



