Modelling Terrestrial Clear-Air Microwave Radio Fading

Abstract

The technology of communication systems between population centres has undergone much change over the last century an a half, but radio links continue to be an important part of communication networks. A challenging part of their design is allowing for variations in received signal level, known as radio fading and enhancement, due to the atmosphere between transmitter and receiver. At high frequencies rain fading is the limiting factor, but below about 10 GHz, temperature and humidity gradients, in the absence of precipitation, may produce clear-air fading that becomes the limiting factor. As the refractive index of the air at radio frequencies depends on temperature and humidity, vertical gradients of these parameters cause bending of ray-paths. Multiple signals may arrive at the receiver over different paths, resulting in multipath fading. Sometimes almost no signal at all is able to find its way from transmitter to receiver, resulting in an impairment known as median depression; this may last for an hour or more, with median signal level up to 50 dB below normal. Recent long-term observations show this fading to be particularly severe in some parts of Australia, but not well predicted by pre-existing models. This thesis develops a new international model for clear-air fading. Weather forecasting has made significant progress in recent years due to numerical weather prediction (NWP) models, so radio propagation researchers have aimed to use these models to predict the state of the atmosphere, and Fourier split-step parabolic equation modelling (PEM) to predict radio propagation. Considering this, we begin this thesis by investigating Fourier split-step PEM, developing new techniques for dealing with finite conductivity lower boundaries, estimating the absorbing upper boundary height, and for dealing with irregular terrain, in both two and three dimensions. A brief description of the internationally adopted empirical model for diffraction over terrain (Rec. ITU-R P526-15, 2019), completes this chapter. We then examine radio refractivity gradient cumulative distributions derived from NWP data, comparing them with measurements from radiosondes, and data from sensors mounted on towers. We find the NWP prediction of anomalous gradients in the surface atmospheric layer to be poor, and develop a new parameter, surface refractivity anomaly, derived from surface weather station time-series data. We find this parameter useful in predicting vertical radio refractivity gradients in the atmospheric surface layer. Due to NWP surface gradient accuracy problems, we adopt the empirical regression model approach to fading severity prediction. This is not new, but we now have the benefit of more fading data from more regions of the world, and we have our new prediction parameters, generated from several years of data from thousands of worldwide weather stations. We make novel refinements to the modelling of clear-air fading, by first replacing ordinary least squares (OLS) regression with generalised least squares (GLS) regression, to take spatial correlation into account. We then employ the geostatistical technique of universal kriging, to further improve prediction accuracy. Our new fading model, as described in this thesis, is now the internationally approved terrestrial line-of-sight model for fading due to multipath and related mechanisms (Rec. ITU-R P.530-18, 2021).