This doctoral dissertation investigates the propagation of electromagnetic waves through the troposphere under clear-sky, cloudy, and precipitating conditions, with particular emphasis on atmospheric attenuation, radiative transfer processes, and tropospheric delay effects relevant to satellite communications and Earth observation systems. The research combines physical modelling, satellite observations, numerical simulations, and machine learning techniques to improve the characterization and prediction of atmospheric phenomena affecting signal propagation from microwave to millimeter-wave frequencies. A first contribution concerns the estimation and validation of Zenith Hydrostatic Delay (ZHD), Zenith Wet Delay (ZWD), and Zenith Total Delay (ZTD) using ERA5 reanalysis data, GNSS observations, and internationally recognized propagation models. Several mapping functions, including Niell, VMF1, GPT3, and Ifadis, are assessed for the reconstruction of slant-path delays, thereby enhancing atmospheric correction capabilities for satellite navigation and communication systems. One of the principal contributions of this dissertation is the development and extension of the Sky-Noise-Eddington Model (SNEM) into its advanced version, SNEM-2 (Statistical Numerical Electromagnetic Model 2). The proposed framework integrates radiative transfer formulations, cloud microphysics, atmospheric thermodynamics, and hydrometeor profiles derived from satellite observations to simulate both atmospheric attenuation and brightness temperature over a broad frequency range extending up to the W-band. Particular attention is devoted to mixed-phase clouds, melting-layer processes, ice hydrometeors, and frequency-dependent scattering mechanisms described through Mie theory and radiative transfer modelling. The study exploits extensive datasets derived from ERA5 reanalysis products, CloudSat observations, the Dual-Frequency Precipitation Radar (DPR) aboard the Global Precipitation Measurement (GPM) mission, Tropical Rainfall Measuring Mission (TRMM) observations, and ground-based measurements. A dedicated satellite-coincidence framework has been developed to integrate active and passive remote-sensing observations and generate physically consistent atmospheric profiles for attenuation analysis and model validation. Furthermore, global attenuation climatologies at Ku-, Ka-, and W-band frequencies have been produced and analysed, highlighting the influence of latitude, seasonal variability, cloud microphysics, and precipitation regimes on propagation losses. Machine learning methodologies, including Long Short-Term Memory (LSTM) neural networks, Generative Adversarial Networks (GANs), and Conditional Generative Adversarial Networks (CGANs), have been investigated as surrogate models for computationally intensive radiative transfer simulations. These approaches enable the emulation of complex atmospheric processes, the reconstruction of realistic synthetic cloud structures, and the acceleration of attenuation prediction while preserving consistency with physically based models. The dissertation further investigates the relationship between atmospheric attenuation, brightness temperature, and thermodynamic properties of the atmosphere, leading to the formulation and validation of predictive attenuation laws obtained through the integration of satellite observations and physically based modelling approaches. The results demonstrate the potential of hybrid methodologies that combine physical principles and artificial intelligence to support the design, optimization, and operation of next-generation satellite communication systems operating in the Ka-, V-, and W-band frequency ranges. Overall, this work contributes to the advancement of atmospheric propagation modelling, radiative transfer simulation, and the design of high-frequency satellite systems by providing novel methodologies and tools for attenuation prediction, atmospheric correction, and the future development of AI-assisted electromagnetic propagation frameworks.
Questa tesi di dottorato analizza la propagazione delle onde elettromagnetiche attraverso la troposfera in condizioni di cielo sereno, nuvolosità e precipitazione, con particolare attenzione all’attenuazione atmosferica, ai processi di trasferimento radiativo e agli effetti del ritardo troposferico rilevanti per le comunicazioni satellitari e i sistemi di osservazione della Terra. La ricerca combina modellazione fisica, osservazioni satellitari, simulazioni numeriche e tecniche di apprendimento automatico al fine di migliorare la caratterizzazione e la previsione dei fenomeni che degradano la propagazione dei segnali dalle microonde alle onde millimetriche. Un primo contributo riguarda la stima e la validazione dello Zenith Hydrostatic Delay (ZHD), dello Zenith Wet Delay (ZWD) e dello Zenith Total Delay (ZTD), utilizzando dati di rianalisi ERA5, osservazioni GNSS e modelli di propagazione internazionalmente riconosciuti. Diverse funzioni di mappatura, tra cui Niell, VMF1, GPT3 e Ifadis, vengono analizzate per la ricostruzione dei ritardi lungo percorsi inclinati (slant path), migliorando le correzioni atmosferiche per i sistemi di navigazione e comunicazione satellitare. Uno dei contributi principali della tesi è rappresentato dallo sviluppo e dall’estensione dello Sky-Noise-Eddington Model (SNEM) nella nuova versione avanzata SNEM-2 (Statistical Numerical Electromagnetic Model 2). Il modello integra formulazioni di trasferimento radiativo, microfisica delle nubi, termodinamica atmosferica e profili di idrometeore derivati da osservazioni satellitari per simulare l’attenuazione e la temperatura di brillanza (Brightness Temperature) su un ampio intervallo di frequenze che si estende fino alla banda W. Particolare attenzione è dedicata alle nubi multifase, ai processi associati al melting layer, agli idrometeori ghiacciati e ai meccanismi di scattering dipendenti dalla frequenza descritti mediante la teoria di Mie e la teoria del trasferimento radiativo. Lo studio utilizza grandi quantità di dati provenienti dalle rianalisi ERA5, dalle osservazioni CloudSat, dai prodotti del Dual-Frequency Precipitation Radar (DPR) della missione GPM, dalle misure TRMM e da osservazioni a terra. È stato sviluppato uno specifico framework di coincidenze satellitari per integrare osservazioni attive e passive e generare profili atmosferici fisicamente consistenti per l’analisi dell’attenuazione e la validazione dei modelli. Sono state inoltre prodotte e analizzate climatologie globali di attenuazione nelle bande Ku, Ka e W, evidenziando l’influenza della latitudine, della stagionalità, della microfisica delle nubi e dei regimi precipitativi sulle perdite di propagazione. Le metodologie di Machine Learning, incluse reti neurali LSTM, Generative Adversarial Networks (GAN) e Conditional Generative Adversarial Networks (CGAN), sono state investigate come modelli surrogati dei codici di trasferimento radiativo. Tali approcci consentono di emulare processi atmosferici complessi, ricostruire strutture nuvolose sintetiche e accelerare la previsione dell’attenuazione mantenendo al contempo la coerenza con i modelli fisici di riferimento. La tesi approfondisce inoltre la relazione tra attenuazione, temperatura di brillanza e proprietà termodinamiche dell’atmosfera, portando alla formulazione e validazione di leggi predittive dell’attenuazione ottenute mediante l’integrazione di osservazioni satellitari e modellazione fisica. I risultati dimostrano il potenziale degli approcci ibridi, che combinano fisica e intelligenza artificiale, per supportare la progettazione e l’ottimizzazione dei sistemi di comunicazione satellitare di nuova generazione operanti nelle bande Ka, V e W. Gli sviluppi futuri includono la convalida sperimentale basata su CubeSat, modelli troposferici potenziati dall'intelligenza artificiale e ulteriori progressi nei quadri di propagazione atmosferica unificati. Nel complesso, questo lavoro contribuisce all’avanzamento della modellazione della propagazione atmosferica, delle simulazioni di trasferimento radiativo e della progettazione di sistemi satellitari ad alta frequenza, fornendo nuovi strumenti e metodologie per la previsione dell’attenuazione, la correzione atmosferica e lo sviluppo di future applicazioni basate sull’intelligenza artificiale nel campo della propagazione elettromagnetica.
Characterization of Radiowave Propagation in the Troposphere under Precipitation Conditions / Cicolin, P.. - (2026 Jun 04).
Characterization of Radiowave Propagation in the Troposphere under Precipitation Conditions
CICOLIN, PAOLO
04/06/2026
Abstract
This doctoral dissertation investigates the propagation of electromagnetic waves through the troposphere under clear-sky, cloudy, and precipitating conditions, with particular emphasis on atmospheric attenuation, radiative transfer processes, and tropospheric delay effects relevant to satellite communications and Earth observation systems. The research combines physical modelling, satellite observations, numerical simulations, and machine learning techniques to improve the characterization and prediction of atmospheric phenomena affecting signal propagation from microwave to millimeter-wave frequencies. A first contribution concerns the estimation and validation of Zenith Hydrostatic Delay (ZHD), Zenith Wet Delay (ZWD), and Zenith Total Delay (ZTD) using ERA5 reanalysis data, GNSS observations, and internationally recognized propagation models. Several mapping functions, including Niell, VMF1, GPT3, and Ifadis, are assessed for the reconstruction of slant-path delays, thereby enhancing atmospheric correction capabilities for satellite navigation and communication systems. One of the principal contributions of this dissertation is the development and extension of the Sky-Noise-Eddington Model (SNEM) into its advanced version, SNEM-2 (Statistical Numerical Electromagnetic Model 2). The proposed framework integrates radiative transfer formulations, cloud microphysics, atmospheric thermodynamics, and hydrometeor profiles derived from satellite observations to simulate both atmospheric attenuation and brightness temperature over a broad frequency range extending up to the W-band. Particular attention is devoted to mixed-phase clouds, melting-layer processes, ice hydrometeors, and frequency-dependent scattering mechanisms described through Mie theory and radiative transfer modelling. The study exploits extensive datasets derived from ERA5 reanalysis products, CloudSat observations, the Dual-Frequency Precipitation Radar (DPR) aboard the Global Precipitation Measurement (GPM) mission, Tropical Rainfall Measuring Mission (TRMM) observations, and ground-based measurements. A dedicated satellite-coincidence framework has been developed to integrate active and passive remote-sensing observations and generate physically consistent atmospheric profiles for attenuation analysis and model validation. Furthermore, global attenuation climatologies at Ku-, Ka-, and W-band frequencies have been produced and analysed, highlighting the influence of latitude, seasonal variability, cloud microphysics, and precipitation regimes on propagation losses. Machine learning methodologies, including Long Short-Term Memory (LSTM) neural networks, Generative Adversarial Networks (GANs), and Conditional Generative Adversarial Networks (CGANs), have been investigated as surrogate models for computationally intensive radiative transfer simulations. These approaches enable the emulation of complex atmospheric processes, the reconstruction of realistic synthetic cloud structures, and the acceleration of attenuation prediction while preserving consistency with physically based models. The dissertation further investigates the relationship between atmospheric attenuation, brightness temperature, and thermodynamic properties of the atmosphere, leading to the formulation and validation of predictive attenuation laws obtained through the integration of satellite observations and physically based modelling approaches. The results demonstrate the potential of hybrid methodologies that combine physical principles and artificial intelligence to support the design, optimization, and operation of next-generation satellite communication systems operating in the Ka-, V-, and W-band frequency ranges. Overall, this work contributes to the advancement of atmospheric propagation modelling, radiative transfer simulation, and the design of high-frequency satellite systems by providing novel methodologies and tools for attenuation prediction, atmospheric correction, and the future development of AI-assisted electromagnetic propagation frameworks.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


