Air quality changes during the COVID-19 pandemic guided by robust virus-spreading data in Italy

This paper aims to assess the impact of restrictive measures against the COVID-19 spread on the air quality of the most representative urban centers in Italy during the 66 days of the first lockdown, integrating a broad and detailed set of socioeconomic and health data into machine learning techniques and correlation analysis. Hierarchical Clustering analysis applied to all 104 Italian provinces indicated a group of six provinces to represent the urban environment in Italy. In contrast, correlation analyses suggested two meteorological parameters and four other air quality parameters as the most skilful at expressing changes in air quality during the first lockdown. Filtering the effects of seasonality, NO concentrations were the ones that most acted in improving urban air quality, showing reductions of up to 48% in all analyzed provinces, directly related to reductions in population mobility in this period (other studies reported an incisive role of pollutants as NO2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$NO_{2}$$\end{document} and PM10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$PM_{10}$$\end{document} or PM2.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$PM_{2.5}$$\end{document} in the SARS-CoV-3 spread). However, there were increases in PM10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$PM_{10}$$\end{document} concentrations related to the use of wood burning for heating, and in SO2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$SO_2$$\end{document} concentrations associated with the food industry, a sector slightly affected by the restrictive measures for being framed as essential. Naples was the only province which reported concentration reductions in all pollutants evaluated, including ozone (7%). However, it was the one that registered the most significant increases during the first days after the lockdown, probably due to the less restrictive measures applied to provinces with the lowest contamination numbers.