An analysis of data from three sets of experiments in transhorizon microwave propagation is presented. In each set, a 3‐Gc/s signal was transmitted over a 163‐km path, and the wave arriving at the receiver was examined using a rapidly scanning phased‐array antenna. In one mode of operation, transmitter and receiver were synchronously swept over a 20‐Mc/s frequency band. The results show considerable variability in all quantities derived from the data.Received power patterns on occasions were scarcely broader than the receiving antenna pattern and at other times showed considerable power arriving off the great‐circle bearing. A classification of such patterns into "stable” and “ unstable” showed no strong association with time of day. Signal level distributions were often, but by no means always, of the Rayleigh form. The time lag required for the signal autocorrelation to drop to 0.5 — indicative of fading period — varied from a half second to a few seconds. A similar measure of transmissable bandwidth — in terms of frequency correlation — showed a variability from 1 or 2 Mc/s to more than 20. The interrelationship between all these quantities, and their connection with meteorological parameters, showed interesting individual relationships but no clear‐cut overall pattern, except for a general increase in signal level associated with pronounced inversions.In comparison with various atmospheric models, it becomes obvious that no single simple model is sufficient. On some occasions the received signal showed characteristics predicted from the theory of incoherent scatter by turbulent eddies: distorted angular response patterns with an average beam broadening, Rayleigh‐type signal level distribution, a modest fading rate, and limited transmissable bandwidth. On other occasions, the records showed evidence of marked atmospheric stability and stratification: unbroadened and undistorted antenna response patterns, a more restricted and non ‐ Rayleigh signal level distribution, variable fading rates, and extremely wide bandwidths. There are still other cases where these associations are not well defined.
Read full abstract