A carregar...
Publicação
Modelling and Prediction of Aerosol Precursors in the CLOUD chamber
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 4.06 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
The CLOUD (Cosmics Leaving Outdoor Droplets) experiment at CERN addresses the largest source of uncertainty in global climate models: the effect of aerosols in cloud formation. This dissertation focused on optimizing data analysis routines on the CLOUD collaboration, aiming to reduce these climate uncertainties. The primary objective was twofold: to enhance the accuracy of the New Particle Formation rate calculation and to develop a robust predictive model for the rate of change of the key aerosol precursor, sulfuric acid. The methodological work involved the re-implementation and optimization of the Formation Rate calculation script in Python, which critically corrected a dimensional inconsistency in the coagulation sink term found in previous implementations, and incorporated advanced signal processing techniques for a more scientifically rigorous result. Furthermore, Machine Learning models were developed using CLOUD16 campaign data to predict the sulfuric acid concentration, incorporating key experimental parameters. After comparison, the non-linear XGBoost model was selected as the superior predictive tool, demonstrating a greater capacity to capture the complex, non-linear relationships driving sulfuric acid dynamics. In conclusion, this work provides a new, reliable analytical and predictive framework for the CLOUD collaboration, contributing directly to more precise data interpretation and supporting a deeper understanding of aerosol formation mechanisms.
Descrição
Tese de Mestrado, Engenharia Física, 2025, Universidade de Lisboa, Faculdade de Ciências
Palavras-chave
Cloud Formation New Particle Formation Machine Learning Aerosols