Abstract: | A model for the statistical prediction of fade durations, applicable both to scintillation and rain-induced effects, is presented in this paper. It assumes that durations longer than about 1 minute are lognormally distributed whereas shorter durations follow a power-law distribution. The model is assessed using data contained in the CCIR data bank and from measurements performed with the SIRIO and Olympus satellites. The model is tested using the same data, and the results give multiplicative errors in the range from 0.6 to 1.8 (averages) and from 0.3 to 3.0 (r.m.s.) in the estimation of the fade durations exceeded for fixed probability. |