Seasonal and diurnal variations of ozone concentration and influencing factors: an application of deep hybrid AI models in Craiova city

Abstract

Abstract The seasonal and diurnal variations in ground-level ozone concentrations and their relationships with other atmospheric components and various meteorological parameters were investigated in Craiova, Romania. By analysing hourly and seasonal trends, this study aims to identify peak ozone formation periods and the key environmental factors influencing their variability. The dataset comprises hourly ozone, air pollutant and meteorological data, enabling a comprehensive assessment of ozone dynamics. The dataset spans five years (2020–2024) and comprises two categories of data: (i) seven meteorological parameters and (ii) twelve air pollutants. The open-source data comes from four air quality monitoring stations located in Craiova. The source of the dataset is the Romanian Environmental Agency’s website ( https://calitateaer.ro/ ). Statistical, machine learning, and deep learning techniques, including time-series decomposition, correlation analysis, regression modelling, and spectral analysis, were employed to extract meaningful insights. The findings could enhance understanding of ozone formation mechanisms, aiding policymakers in designing effective air quality management strategies.