Skip to Main content Skip to Navigation

Etude de la variabilité micro-échelle des précipitations : Application à la propagation des ondes millimétriques en SATCOM

Louis de Montera 1
LATMOS - Laboratoire Atmosphères, Milieux, Observations Spatiales
Abstract : At the EHF band (20-50 GHz), gases, clouds and especially rain provoke an attenuation of the signal between terrestrial telecommunication stations and satellite. Attenuation time series exhibit interesting characteristics, such as highly volatile periods (heteroscedasticity) and fat tailed distributions. Their statistical behavior is similar to some stock exchange rate, which suggests that prediction models originally developed for financial applications could be appropriate. The statistical analysis of attenuation time series leads to a non linear ARIMA/GARCH model. This model obtains a better forecasting performance than the other existing models, in particular because it estimates the prediction error conditional variance.
In order to predict the uplink attenuation from the downlink attenuation that operates at a different frequency, a frequency scaling model has been added to the prediction model. The separation of the attenuation effects (gas, clouds, rain) is performed by a neural network. Then each component is scaled to the required frequency by means of specific scaling factors. The uncertainty of these scaling factors implies a combined treatment of the prediction error and of the error due to frequency scaling.
The prediction model with frequency scaling, developed with measurements of the 20 and 30 GHz beacons of the OLYMPUS satellite, is then tested with recent data from the SYRACUSE propagation experiments. First results concerning log term attenuation statistics are then presented and compared with standard ITU models.
In order to better understand the link between attenuation and its physical causes, an approach based on rain fractal properties is then presented. Indeed, the analogy between rain and finance can be extended, because both phenomena are linked to turbulent processes, and therefore show interesting scale invariance properties. Unfortunately, the multifractal analysis cannot be applied directly to attenuation time series. In a first step, the multifractal analysis is therefore applied to rain rate time series. An assessment of the effect of rain-no rain intermittency on the multifractal analysis shows that it provokes a break in the scaling and may lead to biased parameters. The multifractal analysis is then performed event by event, i.e. with uninterrupted rain periods. The results show that rain can be modeled by a FIF (Fractionally Integrated Flux) which is threholded in order to simulate rain-no rain intermittency.
The multifractal model of rain is then used to simulate an Earth-to-satellite link and to generate synthetic rain attenuation time series. The multifractal analysis of these simulated time series permits to understand why the attenuation process is difficult to model. In particular, although rain fields exhibit a multifractal behavior, attenuation time series do not have stable scale invariance properties and a leveling-off of the power spectrum may even be observed at higher frequencies. These results show that spectrum leveling-off observed in the data is not only due to the presence of a scintillation noise.
Complete list of metadata

Cited literature [108 references]  Display  Hide  Download
Contributor : Louis de Montera Connect in order to contact the contributor
Submitted on : Thursday, May 14, 2009 - 11:04:32 AM
Last modification on : Tuesday, November 16, 2021 - 4:08:25 AM
Long-term archiving on: : Thursday, June 10, 2010 - 11:09:26 PM


  • HAL Id : tel-00384050, version 1


Louis de Montera. Etude de la variabilité micro-échelle des précipitations : Application à la propagation des ondes millimétriques en SATCOM. Géophysique [physics.geo-ph]. Université de Versailles-Saint Quentin en Yvelines, 2008. Français. ⟨tel-00384050⟩



Les métriques sont temporairement indisponibles