Seasonal analysis and prediction of rain-fall effects in Eastern South Africa at microwave frequencies

Network fade countermeasures for link budget can be better implemented based on the knowledge of seasonal variability of rainfall attenuation in a locality. Therefore, in this study, a seasonal approach is applied to estimate the effects of spatial rainfall attenuation in Durban (29°52'S, 30°58'E), South Africa using two-year rainfall data obtained from the RD-80 Joss-Waldvogel (J-W) distrometer. An analysis is undertaken for different seasons to obtain the rainfall rate exceedences at 0.001%, 0.01%, 0.1% and 1% of time. Consequently, rainfall drop-size distribution (DSD) models are developed for the control site at different seasons for the same period. The probability density analysis for each model indicates that the lognormal distribution best fits the summer and autumn season with percentage root-mean-square errors (RMS) of 30% and 26% respectively; gamma distribution fits winter season with RMS error of 16% and Weibull distribution fits spring season with RMS error of 26%. The results from the rainfall rate and rainfall DSD are combined to estimate the rainfall specific attenuation, by applying spherical droplet assumption for Mie scattering techniques, between 2 GHz and 1000 GHz. With this, the seasonal k and α coefficients for specific attenuation are derived from the best rainfall DSD models, using regression technique at 2.5 GHz, 25 GHz, 40 GHz and 100 GHz. At these frequencies, the results show that the predicted specific attenuation coefficients for all seasonal rainfall rates at the control site are lower, when compared to those from ITU-R models. It is concluded that specific attenuation levels may be similar and more intense in summer and autumn seasons, while, lower and less intense in autumn and winter seasons at similar rainfall rates.