We present the NEFOCAST project (named by the contraction of “Nefele”, which is the Italian spelling for the mythological cloud nymph Nephele, and “forecast”), funded by the Tuscany Region, about the feasibility of a system for the detection and monitoring of precipitation fields over the regional territory based on the use of a widespread network of new-generation Eutelsat “SmartLNB” (smart low-noise block converter) domestic terminals. Though primarily intended for interactive satellite services, these devices can also be used as weather sensors, as they have the capability of measuring the rain-induced attenuation incurred by the downlink signal and relaying it on an auxiliary return channel. We illustrate the NEFOCAST system architecture, consisting of the network of ground sensor terminals, the space segment, and the service center, which has the task of processing the information relayed by the terminals for generating rain field maps. We discuss a few methods that allow the conversion of a rain attenuation measurement into an instantaneous rainfall rate. Specifically, we discuss an exponential model relating the specific rain attenuation to the rainfall rate, whose coefficients were obtained from extensive experimental data. The above model permits the inferring of the rainfall rate from the total signal attenuation provided by the SmartLNB and from the link geometry knowledge. Some preliminary results obtained from a SmartLNB installed in Pisa are presented and compared with the output of a conventional tipping bucket rain gauge. It is shown that the NEFOCAST sensor is able to track the fast-varying rainfall rate accurately with no delay, as opposed to a conventional gauge.

In this paper, we present results from the NEFOCAST Project, funded by the Tuscany Region, aiming at detecting and estimating rainfall fields from the opportunistic use of the rain-induced excess attenuation incurred in the downlink channel by a commercial DVB satellite signal. The attenuation is estimated by reverse-engineering the effects of the various propagation phenomena affecting the received signal, among which, in first place, the perturbations factors affecting geostationary orbits, such as the gravitational attraction from the moon and the sun and the inhomogeneity in Earth mass distribution and, second, the small-scale irregularities in the atmospheric refractive index, causing rapid fluctuations in signal amplitude. The latter impairments, in particular, even if periodically counteracted by correction maneuvers, may give rise to significant departures of the actual satellite position from the nominal orbit. A further problem to deal with is the daily and seasonal random fluctuation of the rain height and altitude/size of the associated melting layer. All of the above issues lead to nonnegligible random deviations from the dry nominal downlink attenuation, that can be misinterpreted as rain events. In this paper, we show how to counteract these issues by employing two differentially configured Kalman filters designed to track slow and fast changes of the received signal-to-noise ratio, so that the rain events can be reliably detected and the relevant rainfall rate estimated.

Physical characteristics of precipitation, like temporal and spatial variability, jointly with coverage and costs of conventional meteorological devices for quantitative rainfall estimation (i.e., rain gauges, disdrometers, weather radars) make the precipitation monitoring a complex task. However, real-time rainfall maps are an important tool for many applications, dealing with environment, social activities, and business. Recently, the use of “opportunistic” methods to estimate rainfall has been investigated, highlighting the possibility to exploit inexpensive opportunities to augment information about precipitation. This article deals with smart low-noise blocks (SmartLNBs) converters, which are commercially available interactive digital video broadcasting (DVB) receivers designed to be used as bidirectional modems for commercial interactive TV applications. In the last few years, an algorithm that converts the SmartLNB raw data into attenuation values, from which the rainfall rate is obtained, has been developed and evaluated. The aim of this article is to describe the improvements of the rainfall estimation from SmartLNBs brought by significant changes in the data acquisition from SmartLNB and by algorithms’ update. One year of data collected in Rome and Tuscany (Italy) are analyzed to test the performance of SmartLNB in estimating rainfall accumulation with respect to co-located rain gauges and disdrometer in the new configuration. Comparing SmartLNB and disdrometer data in Rome, we obtained root mean square error (RMSE) equal to 7.3 mm, normalized mean absolute error (NMAE) equal to 51%, with a correlation coefficient of 0.67, that can point out the maturity of the technique.

Accurate precipitation measurement is critical for managing flood and drought risks. Traditional meteorological tools, such as rain gauges and remote sensors, have limitations in resolution, coverage, and cost-effectiveness. Recently, the opportunistic use of microwave communication signals has been explored to improve precipitation estimation. While there is growing interest in using satellite-to-earth microwave links (SMLs) for machine learning-based precipitation estimation, direct rainfall estimation from raw signal-to-noise ratio (SNR) data via deep learning remains underexplored. This study investigates a range of machine learning (ML) approaches, including deep learning (DL) models and traditional methods like gradient boosting machine (GBM), for estimating rainfall rates from SNR data collected by interactive satellite receivers. We develop real-time models for rainfall detection and estimation using downlink SNR signals from satellites to user terminals. By leveraging a year-long dataset from multiple locations—including SNR measurements paired with disdrometer and rain-gauge data—we explore and evaluate various ML models. Our final models include ensemble approaches for both rainfall detection and cumulative rainfall estimation. The proposed models provide a reliable solution for estimating precipitation using Earth–satellite microwave links, potentially improving precipitation monitoring. Compared to the state-of-the-art power-law-based models applied to similar datasets reported in the literature, our ML models achieve a $46\%$ reduction in the root mean squared error (RMSE) for event-based cumulative precipitation predictions.

In this paper we present results from the NEFOCAST project, funded by the Tuscany Region, aiming at detecting and estimating rainfall fields from the opportunistic use of the rain-induced excess attenuation incurred in the downlink channel by a commercial DVB satellite signal. The attenuation is estimated by reverse-engineering the effects of the various propagation phenomena affecting the received signal, among which, in first place, the perturbations factors affecting geostationary orbits, such as the gravitational attraction from the moon and the sun and the inhomogeneity in Earth mass distribution and, secondly, the small-scale irregularities in the atmospheric refractive index, causing rapid fluctuations in signal amplitude. The latter impairments, in particular, even if periodically counteracted by correction maneuvers, may give rise to significant departures of the actual satellite position from the nominal orbit. A further problem to deal with is the daily and seasonal random fluctuation of the rain height and altitude/size of the associated melting layer. All of the above issues lead to non-negligible random deviations from the dry nominal downlink attenuation, that can be misinterpreted as rain events. In this paper we show how to counteract these issues by employing two differentially-configured Kalman filters designed to track slow and fast changes of the received signal-to-noise ratio, so that the rain events can be reliably detected and the relevant rainfall rate estimated.

This study presents a framework for opportunistic rainfall detection and estimation exploiting Signal to Noise Ratio (SNR) data from satellite downlink signals. We develop a heterogeneous ensemble including diverse Machine Learning (ML) architectures for the classification task, namely Conditional Random Field (CRF), Gated Recurrent Unit (GRU), Chronos-Bolt, and TSPulse. Evaluating models characterized by distinct assumptions, scales, and computational complexities provides critical insights into their behavior for the weather and climate remote sensing domain. The resulting ensemble yields a significant improvement in precipitation classification performance. For the regression task, a traditional model leveraging the well-known International Telecommunication Union (ITU) recommendations is compared against a Gradient Boosting Machine (GBM), with the ML-based approach demonstrating a substantial improvement in Key Performance Indicators (KPIs). Finally, we tested the integration of the classification and regression modules via a hard and a soft gating mechanism. While these combinations enhance several point-to-point metrics, they do not improve the estimation of event-total accumulated rainfall. These findings highlight the potential of ML approaches for satellite-based opportunistic environmental monitoring.