Real-Time Rainfall Estimation Using Satellite Signals: Development and Assessment of a New Procedure

Abstract

This contribution presents a comprehensive methodology for the real-time estimation of the rain intensity from downlink satellite signals. The enhanced system leverages on Extremely Randomized Trees Classifiers to automatically perform rainfall detection along Earth-satellite links and successively employs an improved procedure to determine the corresponding slant-path rain attenuation. The latter quantity is then exploited to yield realtime rainfall rate estimates with 1-minute time resolution. The accuracy of the proposed methodology is tested using Ka and Q band propagation data, collected in two different sites (Milan and Madrid) and in the framework of the propagation experiments. The results demonstrate the reliability of the automated rain event detector, as well as a satisfactory accuracy in estimating the slant-path rain attenuation and the point rainfall rate. The accuracy is assessed both on a statistical and on an instantaneous basis through the evaluation of different error figures and by inspection of individual time series.

An Online Training Procedure for Rain Detection Models Applied to Satellite Microwave Links

“…[17] used a support vector machine (SVM) to identify rain/nonrain periods from SML signals and determined the attenuation baseline during rainy periods by adopting an LSTM. [18] trained a randomized trees classifier to assess precipitation presence, and estimated the rain attenuation from the received power signal to compute the rain rate with a 1-minute time resolution. [19] compared four ML algorithms with reduced computational complexity to classify rainy and non-rainy periods based on SML data.…”

Modeling Short-Term Effects of Rain on Satellite Link using Machine Learning

“…The data-driven approach taken by ML algorithm outperforms their conventional counterparts for similar computational complexity and can learn the optimal system architecture using simplistic models. In the domain of satellites, ML has been applied to solve various problems including system capacity [15], precipitation estimation [16] and detection of rain events [17] in the received signal time series.…”