A Rain Attenuation Time-Series Synthesizer Based on a Dirac and Lognormal Distribution

Abstract

In Recommendation ITU-R P.1853-1, a stochastic approach is proposed to generate long-term rain attenuation time series , including rain and no rain periods anywhere in the world. Nevertheless, its dynamic properties have been validated so far from experimental rain attenuation time series collected at mid-latitudes only. In the present paper, an effort is conducted to derive analytically the first- and second-order statistical properties of the ITU rain attenuation time-series synthesizer. It is then shown that the ITU synthesizer does not reproduce the first-order statistics (particularly the rain attenuation cumulative distribution function CDF), however, given as input parameters. It also prevents any rain attenuation correlation function other than exponential to be reproduced, which could be penalizing if a worldwide synthesizer that accounts for the local climatology has to be defined. Therefore, a new rain attenuation time-series synthesizer is proposed. It assumes a mixed Dirac-lognormal modeling of the absolute rain attenuation CDF and relies on a stochastic generation in the Fourier plane. It is then shown analytically that the new synthesizer reproduces much better the first-order statistics given as input parameters and enables any rain attenuation correlation function to be reproduced. The ability of each synthesizer to reproduce absolute rain attenuation CDFs given by Recommendation ITU-R P.618 is finally compared on a worldwide basis. It is then concluded that the new rain attenuation time-series synthesizer reproduces the rain attenuation CDF much better, preserves the rain attenuation dynamics of the current ITU synthesizer for simulations at mid-latitudes, and, if it proves to be necessary for worldwide applications, is able to reproduce any rain attenuation correlation function.